# Installing the libraries with the specified version
!pip install --no-deps tensorflow==2.18.0 scikit-learn==1.3.2 matplotlib===3.8.3 seaborn==0.13.2 numpy==1.26.4 pandas==2.2.2 -q --user --no-warn-script-location

import pandas as pd  # Library for data manipulation and analysis.
import numpy as np   # Fundamental package for scientific computing.
import matplotlib.pyplot as plt  # Plotting library for creating visualizations.
import seaborn as sns #For advanced visualizations.
from sklearn.impute import KNNImputer
from sklearn.model_selection import train_test_split  # Function for splitting datasets for training and testing.
#Imports functions for evaluating the performance of machine learning models
from sklearn.metrics import confusion_matrix, f1_score,accuracy_score, recall_score, precision_score, classification_report
#Imports metrics from
from sklearn import metrics
import time  # Module for time-related operations.

import tensorflow as tf #An end-to-end open source machine learning platform
from tensorflow import keras  # High-level neural networks API for deep learning.
from keras import backend   # Abstraction layer for neural network backend engines.
from keras.models import Sequential  # Model for building NN sequentially.
from keras.layers import Dense,Dropout,BatchNormalization   # for creating fully connected neural network layers.

# uncomment and run the following lines for Google Colab
# from google.colab import drive
# drive.mount('/content/drive')

df_Train = pd.read_csv('Data/Train.csv') # Train Dataset
df_Test = pd.read_csv('Data/Test.csv')   # Test Dataset
pd.set_option('display.width', 300)

print(df_Train.head())
print(df_Test.head())

         V1        V2        V3        V4        V5        V6        V7        V8        V9       V10  ...       V32       V33        V34       V35       V36       V37       V38       V39       V40  Target
0 -4.464606 -4.679129  3.101546  0.506130 -0.221083 -2.032511 -2.910870  0.050714 -1.522351  3.761892  ...  3.059700 -1.690440   2.846296  2.235198  6.667486  0.443809 -2.369169  2.950578 -3.480324       0
1  3.365912  3.653381  0.909671 -1.367528  0.332016  2.358938  0.732600 -4.332135  0.565695 -0.101080  ... -1.795474  3.032780  -2.467514  1.894599 -2.297780 -1.731048  5.908837 -0.386345  0.616242       0
2 -3.831843 -5.824444  0.634031 -2.418815 -1.773827  1.016824 -2.098941 -3.173204 -2.081860  5.392621  ... -0.257101  0.803550   4.086219  2.292138  5.360850  0.351993  2.940021  3.839160 -4.309402       0
3  1.618098  1.888342  7.046143 -1.147285  0.083080 -1.529780  0.207309 -2.493629  0.344926  2.118578  ... -3.584425 -2.577474   1.363769  0.622714  5.550100 -1.526796  0.138853  3.101430 -1.277378       0
4 -0.111440  3.872488 -3.758361 -2.982897  3.792714  0.544960  0.205433  4.848994 -1.854920 -6.220023  ...  8.265896  6.629213 -10.068689  1.222987 -3.229763  1.686909 -2.163896 -3.644622  6.510338       0

[5 rows x 41 columns]
         V1        V2        V3        V4        V5        V6        V7        V8        V9       V10  ...       V32       V33       V34       V35       V36       V37        V38       V39       V40  Target
0 -0.613489 -3.819640  2.202302  1.300420 -1.184929 -4.495964 -1.835817  4.722989  1.206140 -0.341909  ...  2.291204 -5.411388  0.870073  0.574479  4.157191  1.428093 -10.511342  0.454664 -1.448363       0
1  0.389608 -0.512341  0.527053 -2.576776 -1.016766  2.235112 -0.441301 -4.405744 -0.332869  1.966794  ... -2.474936  2.493582  0.315165  2.059288  0.683859 -0.485452   5.128350  1.720744 -1.488235       0
2 -0.874861 -0.640632  4.084202 -1.590454  0.525855 -1.957592 -0.695367  1.347309 -1.732348  0.466500  ... -1.318888 -2.997464  0.459664  0.619774  5.631504  1.323512  -1.752154  1.808302  1.675748       0
3  0.238384  1.458607  4.014528  2.534478  1.196987 -3.117330 -0.924035  0.269493  1.322436  0.702345  ...  3.517918 -3.074085 -0.284220  0.954576  3.029331 -1.367198  -3.412140  0.906000 -2.450889       0
4  5.828225  2.768260 -1.234530  2.809264 -1.641648 -1.406698  0.568643  0.965043  1.918379 -2.774855  ...  1.773841 -1.501573 -2.226702  4.776830 -6.559698 -0.805551  -0.276007 -3.858207 -0.537694       0

[5 rows x 41 columns]

#overview of given train and test dataset for sanity and efficient model creation
data = {'Category':['Train','Test'],
               'Shape (Row * Col)': [df_Train.shape,df_Test.shape ],
               'Contains Null':[df_Train.isnull().sum().sum(),df_Test.isnull().sum().sum()],
               'Column names having Null values': [df_Train.columns[df_Train.isnull().any()].tolist(),df_Test.columns[df_Test.isnull().any()].tolist()]
       }

df_overview = pd.DataFrame(data)
df_overview

df_Train.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 20000 entries, 0 to 19999
Data columns (total 41 columns):
 #   Column  Non-Null Count  Dtype  
---  ------  --------------  -----  
 0   V1      19982 non-null  float64
 1   V2      19982 non-null  float64
 2   V3      20000 non-null  float64
 3   V4      20000 non-null  float64
 4   V5      20000 non-null  float64
 5   V6      20000 non-null  float64
 6   V7      20000 non-null  float64
 7   V8      20000 non-null  float64
 8   V9      20000 non-null  float64
 9   V10     20000 non-null  float64
 10  V11     20000 non-null  float64
 11  V12     20000 non-null  float64
 12  V13     20000 non-null  float64
 13  V14     20000 non-null  float64
 14  V15     20000 non-null  float64
 15  V16     20000 non-null  float64
 16  V17     20000 non-null  float64
 17  V18     20000 non-null  float64
 18  V19     20000 non-null  float64
 19  V20     20000 non-null  float64
 20  V21     20000 non-null  float64
 21  V22     20000 non-null  float64
 22  V23     20000 non-null  float64
 23  V24     20000 non-null  float64
 24  V25     20000 non-null  float64
 25  V26     20000 non-null  float64
 26  V27     20000 non-null  float64
 27  V28     20000 non-null  float64
 28  V29     20000 non-null  float64
 29  V30     20000 non-null  float64
 30  V31     20000 non-null  float64
 31  V32     20000 non-null  float64
 32  V33     20000 non-null  float64
 33  V34     20000 non-null  float64
 34  V35     20000 non-null  float64
 35  V36     20000 non-null  float64
 36  V37     20000 non-null  float64
 37  V38     20000 non-null  float64
 38  V39     20000 non-null  float64
 39  V40     20000 non-null  float64
 40  Target  20000 non-null  int64  
dtypes: float64(40), int64(1)
memory usage: 6.3 MB

df_Test.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 41 columns):
 #   Column  Non-Null Count  Dtype  
---  ------  --------------  -----  
 0   V1      4995 non-null   float64
 1   V2      4994 non-null   float64
 2   V3      5000 non-null   float64
 3   V4      5000 non-null   float64
 4   V5      5000 non-null   float64
 5   V6      5000 non-null   float64
 6   V7      5000 non-null   float64
 7   V8      5000 non-null   float64
 8   V9      5000 non-null   float64
 9   V10     5000 non-null   float64
 10  V11     5000 non-null   float64
 11  V12     5000 non-null   float64
 12  V13     5000 non-null   float64
 13  V14     5000 non-null   float64
 14  V15     5000 non-null   float64
 15  V16     5000 non-null   float64
 16  V17     5000 non-null   float64
 17  V18     5000 non-null   float64
 18  V19     5000 non-null   float64
 19  V20     5000 non-null   float64
 20  V21     5000 non-null   float64
 21  V22     5000 non-null   float64
 22  V23     5000 non-null   float64
 23  V24     5000 non-null   float64
 24  V25     5000 non-null   float64
 25  V26     5000 non-null   float64
 26  V27     5000 non-null   float64
 27  V28     5000 non-null   float64
 28  V29     5000 non-null   float64
 29  V30     5000 non-null   float64
 30  V31     5000 non-null   float64
 31  V32     5000 non-null   float64
 32  V33     5000 non-null   float64
 33  V34     5000 non-null   float64
 34  V35     5000 non-null   float64
 35  V36     5000 non-null   float64
 36  V37     5000 non-null   float64
 37  V38     5000 non-null   float64
 38  V39     5000 non-null   float64
 39  V40     5000 non-null   float64
 40  Target  5000 non-null   int64  
dtypes: float64(40), int64(1)
memory usage: 1.6 MB

df_Train.describe().T

df_Test.describe().T

from typing import Optional, Tuple


def histogram_boxplot(
    data: pd.DataFrame, 
    feature: str, 
    figsize: Tuple[int, int] = (12, 7), 
    kde: bool = False, 
    bins: Optional[int] = None
) -> None:
    """
    Create a combined boxplot and histogram visualization.
    
    Parameters:
    -----------
    data : pd.DataFrame
        Input dataframe containing the data
    feature : str
        Column name to visualize
    figsize : Tuple[int, int], default=(12, 7)
        Figure size as (width, height)
    kde : bool, default=False
        Whether to show kernel density estimation curve
    bins : Optional[int], default=None
        Number of histogram bins
    """
    
    # Create subplot grid with shared x-axis
    fig, (box_ax, hist_ax) = plt.subplots(
        nrows = 2,
        sharex=True,  # Share x-axis between subplots
        gridspec_kw={"height_ratios": (0.25, 0.75)},
        figsize=figsize
    )
    
    # Create boxplot with mean indicator
    sns.boxplot(
        data=data, 
        x=feature, 
        ax=box_ax, 
        showmeans=True, 
        color="blue",
    )
    
    
    # Create histogram with conditional bins parameter
    if bins:
        sns.histplot(
            data=data, 
            x=feature, 
            kde=kde, 
            ax=hist_ax, 
            bins=bins,
            palette="magma"
        )
    else:
        sns.histplot(
            data=data, 
            x=feature, 
            kde=kde, 
            ax=hist_ax
        )
    
    # Add mean line to histogram
    hist_ax.axvline(
        data[feature].mean(), 
        color="violet", 
        linestyle="--"
    )
    
    # Add median line to histogram  
    hist_ax.axvline(
        data[feature].median(), 
        color="black", 
        linestyle="-"
    )
    plt.show()

# function to create labeled barplots


def labeled_barplot(data, feature, perc=False, n=None):
    """
    Barplot with percentage at the top

    data: dataframe
    feature: dataframe column
    perc: whether to display percentages instead of count (default is False)
    n: displays the top n category levels (default is None, i.e., display all levels)
    """

    total = len(data[feature])  # length of the column
    count = data[feature].nunique()
    if n is None:
        plt.figure(figsize=(count + 1, 5))
    else:
        plt.figure(figsize=(n + 1, 5))

    plt.xticks(rotation=90, fontsize=15)
    ax = sns.countplot(
        data=data,
        x=feature,
        hue=feature,
        palette="Paired",
        order=data[feature].value_counts().index[:n].sort_values(),
    )

    for p in ax.patches:
        if perc == True:
            label = "{:.1f}%".format(
                100 * p.get_height() / total
            )  # percentage of each class of the category
        else:
            label = p.get_height()  # count of each level of the category

        x = p.get_x() + p.get_width() / 2  # width of the plot
        y = p.get_height()  # height of the plot

        ax.annotate(
            label,
            (x, y),
            ha="center",
            va="center",
            size=12,
            xytext=(0, 5),
            textcoords="offset points",
        )  # annotate the percentage

    plt.show()  # show the plot

# Iterate through each column (feature) in the training dataframe
for feature in df_Train.columns:
    # For each feature, create both a histogram and boxplot visualization
    # Parameters: dataframe, feature name, figure size (8,6), show plot (True), and no custom bins (None)
    histogram_boxplot(df_Train, feature, (8, 4), True, None)

# Extract key observations from the plots
print(f"Observations for {feature}:")
print(f"Distribution shape: {'Normal' if df_Train[feature].skew() < 0.5 else 'Skewed'}")
print(f"Outliers present: {'Yes' if abs(df_Train[feature].max() - df_Train[feature].mean()) > 3*df_Train[feature].std() else 'No'}")
print(f"Mean value: {df_Train[feature].mean():.2f}")
print(f"Data range: {df_Train[feature].min():.2f} to {df_Train[feature].max():.2f}")

Observations for Target:
Distribution shape: Skewed
Outliers present: Yes
Mean value: 0.06
Data range: 0.00 to 1.00

labeled_barplot(df_Train,'Target',perc =True)

labeled_barplot(df_Test,'Target',perc =True)

#performing Bivariate analysis 
cols_list = df_Train.select_dtypes(include=np.number).columns.tolist()
cols_list.remove("Target")
plt.figure(figsize=(30,20))
sns.heatmap(df_Train[cols_list].corr(),annot = True,vmin=-1,vmax =1,fmt = ".2f",cmap = "Spectral")
plt.show()

plt.figure(figsize=(8, 6))
sns.regplot(y=df_Train.V7, x=df_Train.V15)

<Axes: xlabel='V15', ylabel='V7'>

plt.figure(figsize=(8, 6))
sns.regplot(y=df_Train.V2, x=df_Train.V14)

<Axes: xlabel='V14', ylabel='V2'>

plt.figure(figsize=(8, 6))
sns.regplot(y=df_Train.V20, x=df_Train.V6)

<Axes: xlabel='V6', ylabel='V20'>

##Changing Traget Varibale Datatype from Int to float 

df_Train['Target'] = df_Train['Target'].astype(float)
df_Test['Target'] = df_Test['Target'].astype(float)

# Separating dataset for taget variable and spliting in Train ,val and Test ,to avoid data leakage

X = df_Train.drop(['Target'],axis =1)
y = df_Train['Target']

x_train,x_val,y_train,y_val = train_test_split(
    X,y,test_size=0.2,random_state=1,stratify=y)

print(x_train.shape)
print(x_val.shape)
print(y_train.shape)
print(y_val.shape)

(16000, 40)
(4000, 40)
(16000,)
(4000,)

# Dividing test data into X_test and y_test
X_test = df_Test.drop(columns = ['Target'] , axis= 1)
y_test = df_Test["Target"]

from sklearn.impute import SimpleImputer
imputer = SimpleImputer(strategy="median")
impute = imputer.fit(x_train)

x_train_si = impute.transform(x_train)
x_val_si = imputer.transform(x_val)

x_train = pd.DataFrame(x_train_si,columns=x_train.columns, index=x_train.index)
x_val = pd.DataFrame(x_val_si, columns=x_val.columns, index=x_val.index)

def plot(history, name):
    """
    Function to plot loss/accuracy

    history: an object which stores the metrics and losses.
    name: can be one of Loss or Accuracy
    """
    fig, ax = plt.subplots() #Creating a subplot with figure and axes.
    plt.plot(history.history[name]) #Plotting the train accuracy or train loss
    plt.plot(history.history['val_'+name]) #Plotting the validation accuracy or validation loss

    plt.title('Model ' + name.capitalize()) #Defining the title of the plot.
    plt.ylabel(name.capitalize()) #Capitalizing the first letter.
    plt.xlabel('Epoch') #Defining the label for the x-axis.
    fig.legend(['Train', 'Validation'], loc="outside right upper") #Defining the legend, loc controls the position of the legend.

# defining a function to compute different metrics to check performance of a classification model built using statsmodels
def model_performance_classification(
    model, predictors, target, threshold=0.5
):
    """
    Function to compute different metrics to check classification model performance

    model: classifier
    predictors: independent variables
    target: dependent variable
    threshold: threshold for classifying the observation as class 1
    """

    # checking which probabilities are greater than threshold
    pred = model.predict(predictors) > threshold
    # pred_temp = model.predict(predictors) > threshold
    # # rounding off the above values to get classes
    # pred = np.round(pred_temp)

    acc = accuracy_score(target, pred)  # to compute Accuracy
    recall = recall_score(target, pred, average='macro')  # to compute Recall
    precision = precision_score(target, pred, average='macro')  # to compute Precision
    f1 = f1_score(target, pred, average='macro')  # to compute F1-score

    # creating a dataframe of metrics
    df_perf = pd.DataFrame(
        {"Accuracy": acc, "Recall": recall, "Precision": precision, "F1 Score": f1,}, index = [0]
    )

    return df_perf

def make_confusion_matrix(cf,
                          group_names=None,
                          categories='auto',
                          count=True,
                          percent=True,
                          cbar=True,
                          xyticks=True,
                          xyplotlabels=True,
                          sum_stats=True,
                          figsize=None,
                          cmap='Blues',
                          title=None):
    '''
    This function will make a pretty plot of an sklearn Confusion Matrix cm using a Seaborn heatmap visualization.
    Arguments
    '''


    # CODE TO GENERATE TEXT INSIDE EACH SQUARE
    blanks = ['' for i in range(cf.size)]

    if group_names and len(group_names)==cf.size:
        group_labels = ["{}\n".format(value) for value in group_names]
    else:
        group_labels = blanks

    if count:
        group_counts = ["{0:0.0f}\n".format(value) for value in cf.flatten()]
    else:
        group_counts = blanks

    if percent:
        group_percentages = ["{0:.2%}".format(value) for value in cf.flatten()/np.sum(cf)]
    else:
        group_percentages = blanks

    box_labels = [f"{v1}{v2}{v3}".strip() for v1, v2, v3 in zip(group_labels,group_counts,group_percentages)]
    box_labels = np.asarray(box_labels).reshape(cf.shape[0],cf.shape[1])


    # CODE TO GENERATE SUMMARY STATISTICS & TEXT FOR SUMMARY STATS
    if sum_stats:
        #Accuracy is sum of diagonal divided by total observations
        accuracy  = np.trace(cf) / float(np.sum(cf))

        #if it is a binary confusion matrix, show some more stats
        if len(cf)==2:
            #Metrics for Binary Confusion Matrices
            precision = cf[1,1] / sum(cf[:,1])
            recall    = cf[1,1] / sum(cf[1,:])
            f1_score  = 2*precision*recall / (precision + recall)
            stats_text = "\n\nAccuracy={:0.3f}\nPrecision={:0.3f}\nRecall={:0.3f}\nF1 Score={:0.3f}".format(
                accuracy,precision,recall,f1_score)
        else:
            stats_text = "\n\nAccuracy={:0.3f}".format(accuracy)
    else:
        stats_text = ""


    # SET FIGURE PARAMETERS ACCORDING TO OTHER ARGUMENTS
    if figsize==None:
        #Get default figure size if not set
        figsize = plt.rcParams.get('figure.figsize')

    if xyticks==False:
        #Do not show categories if xyticks is False
        categories=False


    # MAKE THE HEATMAP VISUALIZATION
    plt.figure(figsize=figsize)
    sns.heatmap(cf,annot=box_labels,fmt="",cmap=cmap,cbar=cbar,xticklabels=categories,yticklabels=categories)

    if xyplotlabels:
        plt.ylabel('True label')
        plt.xlabel('Predicted label' + stats_text)
    else:
        plt.xlabel(stats_text)
    
    if title:
        plt.title(title)

#Defining the columns of the dataframe which are nothing but the hyper parameters and the metrics.
columns = ["# hidden layers","# neurons - hidden layer","activation function - hidden layer ","# epochs","batch size","optimizer","learning rate, momentum","weight initializer","regularization","train loss","validation loss","train recall","validation recall","time (secs)"]

#Creating a pandas dataframe.
results = pd.DataFrame(columns=columns)

#defining batch_size and epochs to be used as same values for all models 
batch_size = 32
epochs = 50

# clears the current Keras session, resetting all layers and models previously created, freeing up memory and resources.
tf.keras.backend.clear_session()

# initializing the neural network
model_0 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_0.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_0.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_0.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 1)              │           129 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 5,377 (21.00 KB)

 Trainable params: 5,377 (21.00 KB)

 Non-trainable params: 0 (0.00 B)

optimizer = keras.optimizers.SGD()    # defining SGD as the optimizer to be used
model_0.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_0.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.2355 - loss: 0.2048 - val_Recall: 0.5991 - val_loss: 0.0991
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.5939 - loss: 0.0898 - val_Recall: 0.7477 - val_loss: 0.0871
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.7078 - loss: 0.0739 - val_Recall: 0.7568 - val_loss: 0.0779
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.7347 - loss: 0.0683 - val_Recall: 0.7568 - val_loss: 0.0749
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.7788 - loss: 0.0615 - val_Recall: 0.7793 - val_loss: 0.0720
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 257us/step - Recall: 0.7641 - loss: 0.0622 - val_Recall: 0.7883 - val_loss: 0.0700
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.7742 - loss: 0.0620 - val_Recall: 0.7973 - val_loss: 0.0682
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 256us/step - Recall: 0.8101 - loss: 0.0554 - val_Recall: 0.8063 - val_loss: 0.0677
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 258us/step - Recall: 0.7949 - loss: 0.0597 - val_Recall: 0.7973 - val_loss: 0.0661
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 258us/step - Recall: 0.8130 - loss: 0.0577 - val_Recall: 0.8018 - val_loss: 0.0668
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8152 - loss: 0.0582 - val_Recall: 0.8063 - val_loss: 0.0660
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8387 - loss: 0.0501 - val_Recall: 0.8153 - val_loss: 0.0646
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 306us/step - Recall: 0.8316 - loss: 0.0526 - val_Recall: 0.8108 - val_loss: 0.0640
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8111 - loss: 0.0542 - val_Recall: 0.8153 - val_loss: 0.0641
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8273 - loss: 0.0478 - val_Recall: 0.8108 - val_loss: 0.0635
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8344 - loss: 0.0507 - val_Recall: 0.8153 - val_loss: 0.0636
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8123 - loss: 0.0543 - val_Recall: 0.8198 - val_loss: 0.0630
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8406 - loss: 0.0469 - val_Recall: 0.8153 - val_loss: 0.0622
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8246 - loss: 0.0515 - val_Recall: 0.8198 - val_loss: 0.0622
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8584 - loss: 0.0460 - val_Recall: 0.8198 - val_loss: 0.0623
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8641 - loss: 0.0444 - val_Recall: 0.8243 - val_loss: 0.0618
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8286 - loss: 0.0502 - val_Recall: 0.8243 - val_loss: 0.0629
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8323 - loss: 0.0498 - val_Recall: 0.8243 - val_loss: 0.0615
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8354 - loss: 0.0509 - val_Recall: 0.8198 - val_loss: 0.0616
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8384 - loss: 0.0511 - val_Recall: 0.8288 - val_loss: 0.0620
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8635 - loss: 0.0463 - val_Recall: 0.8243 - val_loss: 0.0621
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8417 - loss: 0.0466 - val_Recall: 0.8288 - val_loss: 0.0614
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8544 - loss: 0.0473 - val_Recall: 0.8198 - val_loss: 0.0625
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8629 - loss: 0.0399 - val_Recall: 0.8333 - val_loss: 0.0610
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8521 - loss: 0.0476 - val_Recall: 0.8288 - val_loss: 0.0613
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8441 - loss: 0.0468 - val_Recall: 0.8243 - val_loss: 0.0612
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8637 - loss: 0.0422 - val_Recall: 0.8423 - val_loss: 0.0609
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 261us/step - Recall: 0.8505 - loss: 0.0426 - val_Recall: 0.8378 - val_loss: 0.0609
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 260us/step - Recall: 0.8668 - loss: 0.0401 - val_Recall: 0.8514 - val_loss: 0.0609
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8545 - loss: 0.0456 - val_Recall: 0.8333 - val_loss: 0.0604
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8376 - loss: 0.0485 - val_Recall: 0.8333 - val_loss: 0.0620
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8485 - loss: 0.0434 - val_Recall: 0.8378 - val_loss: 0.0611
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8622 - loss: 0.0402 - val_Recall: 0.8514 - val_loss: 0.0606
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8643 - loss: 0.0406 - val_Recall: 0.8468 - val_loss: 0.0602
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8657 - loss: 0.0418 - val_Recall: 0.8378 - val_loss: 0.0608
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8345 - loss: 0.0493 - val_Recall: 0.8468 - val_loss: 0.0604
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 258us/step - Recall: 0.8530 - loss: 0.0422 - val_Recall: 0.8423 - val_loss: 0.0607
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 258us/step - Recall: 0.8613 - loss: 0.0439 - val_Recall: 0.8423 - val_loss: 0.0603
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 258us/step - Recall: 0.8672 - loss: 0.0401 - val_Recall: 0.8423 - val_loss: 0.0602
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8623 - loss: 0.0410 - val_Recall: 0.8378 - val_loss: 0.0603
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8661 - loss: 0.0419 - val_Recall: 0.8514 - val_loss: 0.0606
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 309us/step - Recall: 0.8708 - loss: 0.0391 - val_Recall: 0.8514 - val_loss: 0.0608
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 262us/step - Recall: 0.8815 - loss: 0.0373 - val_Recall: 0.8423 - val_loss: 0.0610
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8732 - loss: 0.0391 - val_Recall: 0.8468 - val_loss: 0.0601
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 259us/step - Recall: 0.8530 - loss: 0.0414 - val_Recall: 0.8514 - val_loss: 0.0599

print("Time taken in seconds ",end-start)

Time taken in seconds  6.92151403427124

plot(history,'loss')

plot(history,'Recall')

model_0_train_perf = model_performance_classification(model_0, x_train, y_train) #threshold = 0.5
print("Model 0 Train Performance")
print(model_0_train_perf)
print("-"*50)
print("Model 0 Validation Performance")
model_0_val_perf = model_performance_classification(model_0,x_val,y_val)
print(model_0_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 135us/step
Model 0 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.992125  0.934884   0.989206  0.960243
--------------------------------------------------
Model 0 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 155us/step
   Accuracy    Recall  Precision  F1 Score
0      0.99  0.924749   0.977805  0.949515

#let's get classification report on y train and y val for Model 0
y_train_pred_0 = model_0.predict(x_train)
y_val_pred_0 = model_0.predict(x_val)
print("Classification Report - Train data Model_0",end="\n\n")
cr_train_model_0 = classification_report(y_train,y_train_pred_0>0.5)
print(cr_train_model_0)
print("Classification Report - Validation data Model_0",end="\n\n")
cr_val_model_0 = classification_report(y_val,y_val_pred_0>0.5)
print(cr_val_model_0)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 138us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 159us/step
Classification Report - Train data Model_0

              precision    recall  f1-score   support

           0       0.99      1.00      1.00     15112
           1       0.99      0.87      0.92       888

    accuracy                           0.99     16000
   macro avg       0.99      0.93      0.96     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_0

              precision    recall  f1-score   support

           0       0.99      1.00      0.99      3778
           1       0.96      0.85      0.90       222

    accuracy                           0.99      4000
   macro avg       0.98      0.92      0.95      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_0.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 148us/step

#assigning last epoch loss and recall for both train and val for quick comparision
results.loc[0] = [1,[128],
                  ["relu"],50,32,["sgd"] ,
                  [0.01, "-"],"-","-",
                  history.history["loss"][-1]
                  ,history.history["val_loss"][-1]
                  ,history.history["Recall"][-1]
                  ,history.history["val_Recall"][-1],round(end-start,2)]

results.head(1)

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_1 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_1.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_1.add(Dense(64, activation = 'relu'))
model_1.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_1.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 13,569 (53.00 KB)

 Trainable params: 13,569 (53.00 KB)

 Non-trainable params: 0 (0.00 B)

#SGD Optiomizer
optimizer = keras.optimizers.SGD()    # defining SGD as the optimizer to be used
model_1.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_1.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 404us/step - Recall: 0.3526 - loss: 0.1638 - val_Recall: 0.6622 - val_loss: 0.0894
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 290us/step - Recall: 0.6919 - loss: 0.0810 - val_Recall: 0.7477 - val_loss: 0.0772
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.7575 - loss: 0.0723 - val_Recall: 0.7658 - val_loss: 0.0710
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 289us/step - Recall: 0.8019 - loss: 0.0583 - val_Recall: 0.8108 - val_loss: 0.0678
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8150 - loss: 0.0574 - val_Recall: 0.8108 - val_loss: 0.0651
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8153 - loss: 0.0534 - val_Recall: 0.8153 - val_loss: 0.0630
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8270 - loss: 0.0528 - val_Recall: 0.8153 - val_loss: 0.0617
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8386 - loss: 0.0499 - val_Recall: 0.8198 - val_loss: 0.0610
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 293us/step - Recall: 0.8429 - loss: 0.0472 - val_Recall: 0.8198 - val_loss: 0.0605
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 298us/step - Recall: 0.8511 - loss: 0.0463 - val_Recall: 0.8243 - val_loss: 0.0593
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 290us/step - Recall: 0.8486 - loss: 0.0450 - val_Recall: 0.8153 - val_loss: 0.0593
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8399 - loss: 0.0487 - val_Recall: 0.8333 - val_loss: 0.0582
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 289us/step - Recall: 0.8537 - loss: 0.0456 - val_Recall: 0.8378 - val_loss: 0.0581
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8578 - loss: 0.0414 - val_Recall: 0.8468 - val_loss: 0.0571
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 289us/step - Recall: 0.8607 - loss: 0.0424 - val_Recall: 0.8333 - val_loss: 0.0573
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8760 - loss: 0.0380 - val_Recall: 0.8514 - val_loss: 0.0570
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 289us/step - Recall: 0.8895 - loss: 0.0366 - val_Recall: 0.8378 - val_loss: 0.0565
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 292us/step - Recall: 0.8814 - loss: 0.0387 - val_Recall: 0.8514 - val_loss: 0.0562
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8683 - loss: 0.0407 - val_Recall: 0.8559 - val_loss: 0.0567
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8870 - loss: 0.0393 - val_Recall: 0.8468 - val_loss: 0.0562
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8888 - loss: 0.0339 - val_Recall: 0.8468 - val_loss: 0.0561
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.8764 - loss: 0.0396 - val_Recall: 0.8423 - val_loss: 0.0558
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.8821 - loss: 0.0355 - val_Recall: 0.8468 - val_loss: 0.0564
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8817 - loss: 0.0371 - val_Recall: 0.8514 - val_loss: 0.0558
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8818 - loss: 0.0380 - val_Recall: 0.8514 - val_loss: 0.0551
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.8860 - loss: 0.0340 - val_Recall: 0.8514 - val_loss: 0.0559
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.8907 - loss: 0.0323 - val_Recall: 0.8514 - val_loss: 0.0553
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 297us/step - Recall: 0.9005 - loss: 0.0333 - val_Recall: 0.8604 - val_loss: 0.0547
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 291us/step - Recall: 0.8988 - loss: 0.0340 - val_Recall: 0.8604 - val_loss: 0.0550
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8820 - loss: 0.0362 - val_Recall: 0.8468 - val_loss: 0.0551
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 285us/step - Recall: 0.9074 - loss: 0.0306 - val_Recall: 0.8468 - val_loss: 0.0554
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 288us/step - Recall: 0.8976 - loss: 0.0311 - val_Recall: 0.8604 - val_loss: 0.0553
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8905 - loss: 0.0350 - val_Recall: 0.8514 - val_loss: 0.0555
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8894 - loss: 0.0328 - val_Recall: 0.8559 - val_loss: 0.0545
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8823 - loss: 0.0360 - val_Recall: 0.8559 - val_loss: 0.0547
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 285us/step - Recall: 0.8847 - loss: 0.0324 - val_Recall: 0.8559 - val_loss: 0.0557
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 291us/step - Recall: 0.8976 - loss: 0.0298 - val_Recall: 0.8559 - val_loss: 0.0550
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.8975 - loss: 0.0295 - val_Recall: 0.8604 - val_loss: 0.0552
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 296us/step - Recall: 0.8765 - loss: 0.0359 - val_Recall: 0.8514 - val_loss: 0.0553
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.9074 - loss: 0.0296 - val_Recall: 0.8604 - val_loss: 0.0548
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.8925 - loss: 0.0327 - val_Recall: 0.8694 - val_loss: 0.0563
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step - Recall: 0.9039 - loss: 0.0293 - val_Recall: 0.8559 - val_loss: 0.0563
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.8966 - loss: 0.0304 - val_Recall: 0.8604 - val_loss: 0.0564
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 295us/step - Recall: 0.9015 - loss: 0.0298 - val_Recall: 0.8604 - val_loss: 0.0544
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8963 - loss: 0.0285 - val_Recall: 0.8604 - val_loss: 0.0545
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 286us/step - Recall: 0.9091 - loss: 0.0282 - val_Recall: 0.8604 - val_loss: 0.0554
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 285us/step - Recall: 0.8937 - loss: 0.0310 - val_Recall: 0.8694 - val_loss: 0.0560
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 285us/step - Recall: 0.9005 - loss: 0.0304 - val_Recall: 0.8514 - val_loss: 0.0551
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 287us/step - Recall: 0.8933 - loss: 0.0287 - val_Recall: 0.8559 - val_loss: 0.0563
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 285us/step - Recall: 0.8980 - loss: 0.0263 - val_Recall: 0.8694 - val_loss: 0.0552

print("Time taken in seconds ",end-start)

Time taken in seconds  7.699712038040161

plot(history,'loss')

plot(history,'Recall')

model_1_train_perf = model_performance_classification(model_1, x_train, y_train) #threshold = 0.5
print("Model 1 Train Performance")
print(model_1_train_perf)
print("-"*50)
print("/nModel 1 Validation Performance")
model_1_val_perf = model_performance_classification(model_1,x_val,y_val)
print(model_1_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 136us/step
Model 1 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.994437  0.953597   0.992995   0.97237
--------------------------------------------------
/nModel 1 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 159us/step
   Accuracy    Recall  Precision  F1 Score
0   0.99175  0.934155   0.986035  0.958444

#let's get classification report on y train and y val for Model 0
y_train_pred_1 = model_1.predict(x_train)
y_val_pred_1 = model_1.predict(x_val)
print("Classification Report - Train data Model_1",end="\n\n")
cr_train_model_1 = classification_report(y_train,y_train_pred_1>0.5)
print(cr_train_model_1)
print("Classification Report - Validation data Model_1",end="\n\n")
cr_val_model_1 = classification_report(y_val,y_val_pred_1>0.5)
print(cr_val_model_1)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 147us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 171us/step
Classification Report - Train data Model_1

              precision    recall  f1-score   support

           0       0.99      1.00      1.00     15112
           1       0.99      0.91      0.95       888

    accuracy                           0.99     16000
   macro avg       0.99      0.95      0.97     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_1

              precision    recall  f1-score   support

           0       0.99      1.00      1.00      3778
           1       0.98      0.87      0.92       222

    accuracy                           0.99      4000
   macro avg       0.99      0.93      0.96      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_1.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 170us/step

results.loc[1] = [2,[128,64],["relu","relu"],50,32,"sgd",[0.01, "-"],"-","-",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_2 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_2.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_2.add(Dropout(0.5))  # Adding 50% Dropout after first hidden layer
model_2.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_2.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_2.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 13,569 (53.00 KB)

 Trainable params: 13,569 (53.00 KB)

 Non-trainable params: 0 (0.00 B)

#SGD Optiomizer
optimizer = keras.optimizers.SGD()    # defining SGD as the optimizer to be used
model_2.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_2.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 415us/step - Recall: 0.2015 - loss: 0.2092 - val_Recall: 0.5315 - val_loss: 0.1075
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.4688 - loss: 0.1357 - val_Recall: 0.6757 - val_loss: 0.0923
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.5540 - loss: 0.1116 - val_Recall: 0.7342 - val_loss: 0.0838
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.5992 - loss: 0.1044 - val_Recall: 0.7523 - val_loss: 0.0803
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.6374 - loss: 0.0981 - val_Recall: 0.7613 - val_loss: 0.0769
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.6628 - loss: 0.0921 - val_Recall: 0.7658 - val_loss: 0.0742
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.6955 - loss: 0.0880 - val_Recall: 0.7748 - val_loss: 0.0713
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.7119 - loss: 0.0796 - val_Recall: 0.7838 - val_loss: 0.0701
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.7167 - loss: 0.0740 - val_Recall: 0.7838 - val_loss: 0.0688
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7067 - loss: 0.0772 - val_Recall: 0.7883 - val_loss: 0.0670
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.7361 - loss: 0.0788 - val_Recall: 0.7973 - val_loss: 0.0658
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.7508 - loss: 0.0683 - val_Recall: 0.7973 - val_loss: 0.0645
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7452 - loss: 0.0774 - val_Recall: 0.8108 - val_loss: 0.0642
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7685 - loss: 0.0701 - val_Recall: 0.8153 - val_loss: 0.0630
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7562 - loss: 0.0732 - val_Recall: 0.8153 - val_loss: 0.0631
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.7654 - loss: 0.0734 - val_Recall: 0.8198 - val_loss: 0.0629
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.7698 - loss: 0.0634 - val_Recall: 0.8288 - val_loss: 0.0617
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 302us/step - Recall: 0.7775 - loss: 0.0683 - val_Recall: 0.8333 - val_loss: 0.0611
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.7948 - loss: 0.0621 - val_Recall: 0.8333 - val_loss: 0.0608
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.7857 - loss: 0.0641 - val_Recall: 0.8333 - val_loss: 0.0608
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.7854 - loss: 0.0645 - val_Recall: 0.8378 - val_loss: 0.0602
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7752 - loss: 0.0607 - val_Recall: 0.8333 - val_loss: 0.0604
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8119 - loss: 0.0536 - val_Recall: 0.8333 - val_loss: 0.0592
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.7860 - loss: 0.0564 - val_Recall: 0.8423 - val_loss: 0.0595
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7968 - loss: 0.0665 - val_Recall: 0.8423 - val_loss: 0.0585
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8234 - loss: 0.0550 - val_Recall: 0.8378 - val_loss: 0.0583
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.7868 - loss: 0.0623 - val_Recall: 0.8288 - val_loss: 0.0592
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8350 - loss: 0.0502 - val_Recall: 0.8514 - val_loss: 0.0578
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 308us/step - Recall: 0.8165 - loss: 0.0613 - val_Recall: 0.8514 - val_loss: 0.0577
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 300us/step - Recall: 0.8129 - loss: 0.0591 - val_Recall: 0.8514 - val_loss: 0.0577
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 301us/step - Recall: 0.8205 - loss: 0.0601 - val_Recall: 0.8514 - val_loss: 0.0572
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 314us/step - Recall: 0.8162 - loss: 0.0543 - val_Recall: 0.8514 - val_loss: 0.0573
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 310us/step - Recall: 0.8093 - loss: 0.0528 - val_Recall: 0.8514 - val_loss: 0.0573
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8224 - loss: 0.0591 - val_Recall: 0.8468 - val_loss: 0.0578
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 306us/step - Recall: 0.8402 - loss: 0.0457 - val_Recall: 0.8514 - val_loss: 0.0567
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8131 - loss: 0.0549 - val_Recall: 0.8514 - val_loss: 0.0570
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 306us/step - Recall: 0.7975 - loss: 0.0606 - val_Recall: 0.8468 - val_loss: 0.0568
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8121 - loss: 0.0497 - val_Recall: 0.8604 - val_loss: 0.0561
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 353us/step - Recall: 0.8261 - loss: 0.0503 - val_Recall: 0.8604 - val_loss: 0.0567
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 307us/step - Recall: 0.8445 - loss: 0.0490 - val_Recall: 0.8559 - val_loss: 0.0565
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8525 - loss: 0.0468 - val_Recall: 0.8514 - val_loss: 0.0565
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8366 - loss: 0.0543 - val_Recall: 0.8559 - val_loss: 0.0561
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8124 - loss: 0.0597 - val_Recall: 0.8604 - val_loss: 0.0558
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8115 - loss: 0.0539 - val_Recall: 0.8514 - val_loss: 0.0559
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8481 - loss: 0.0525 - val_Recall: 0.8604 - val_loss: 0.0557
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 303us/step - Recall: 0.8656 - loss: 0.0445 - val_Recall: 0.8604 - val_loss: 0.0553
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8173 - loss: 0.0556 - val_Recall: 0.8514 - val_loss: 0.0556
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8575 - loss: 0.0487 - val_Recall: 0.8559 - val_loss: 0.0554
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 304us/step - Recall: 0.8458 - loss: 0.0500 - val_Recall: 0.8604 - val_loss: 0.0557
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 305us/step - Recall: 0.8193 - loss: 0.0535 - val_Recall: 0.8559 - val_loss: 0.0561

print("Time taken in seconds ",end-start)

Time taken in seconds  8.07505202293396

plot(history,'loss')

plot(history,'Recall')

model_2_train_perf = model_performance_classification(model_2, x_train, y_train) #threshold = 0.5
print("Model 2 Train Performance")
print(model_2_train_perf)
print("-"*50)
print("Model 2 Validation Performance")
model_2_val_perf = model_performance_classification(model_2,x_val,y_val)
print(model_2_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 138us/step
Model 2 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.992938  0.941144   0.990906  0.964526
--------------------------------------------------
Model 2 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 162us/step
   Accuracy    Recall  Precision  F1 Score
0   0.99025  0.927002   0.978026  0.950888

#let's get classification report on y train and y val for Model 2
y_train_pred_2 = model_2.predict(x_train)
y_val_pred_2 = model_2.predict(x_val)
print("Classification Report - Train data Model_2",end="\n\n")
cr_train_model_2 = classification_report(y_train,y_train_pred_2>0.5)
print(cr_train_model_2)
print("Classification Report - Validation data Model_2",end="\n\n")
cr_val_model_2 = classification_report(y_val,y_val_pred_2>0.5)
print(cr_val_model_2)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 147us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 169us/step
Classification Report - Train data Model_2

              precision    recall  f1-score   support

           0       0.99      1.00      1.00     15112
           1       0.99      0.88      0.93       888

    accuracy                           0.99     16000
   macro avg       0.99      0.94      0.96     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_2

              precision    recall  f1-score   support

           0       0.99      1.00      0.99      3778
           1       0.96      0.86      0.91       222

    accuracy                           0.99      4000
   macro avg       0.98      0.93      0.95      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_2.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 175us/step

results.loc[2] = [2,[128,64],["relu","relu"],50,32,"sgd",[0.01, "-"],"-","dropout(0.5)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_3 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_3.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_3.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_3.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_3.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_3.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_3.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#SGD Optiomizer
optimizer = keras.optimizers.SGD()    # defining SGD as the optimizer to be used
model_3.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_3.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 435us/step - Recall: 0.1136 - loss: 0.1984 - val_Recall: 0.6081 - val_loss: 0.1010
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 326us/step - Recall: 0.5237 - loss: 0.1175 - val_Recall: 0.7072 - val_loss: 0.0837
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.6236 - loss: 0.0951 - val_Recall: 0.7793 - val_loss: 0.0762
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.6721 - loss: 0.0903 - val_Recall: 0.7658 - val_loss: 0.0713
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.7158 - loss: 0.0804 - val_Recall: 0.8108 - val_loss: 0.0673
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.7613 - loss: 0.0781 - val_Recall: 0.8063 - val_loss: 0.0648
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.7355 - loss: 0.0670 - val_Recall: 0.8063 - val_loss: 0.0624
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.7814 - loss: 0.0717 - val_Recall: 0.8243 - val_loss: 0.0601
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.7632 - loss: 0.0678 - val_Recall: 0.8063 - val_loss: 0.0613
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8116 - loss: 0.0616 - val_Recall: 0.8198 - val_loss: 0.0592
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.7724 - loss: 0.0706 - val_Recall: 0.8423 - val_loss: 0.0572
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8221 - loss: 0.0584 - val_Recall: 0.8288 - val_loss: 0.0560
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.8261 - loss: 0.0527 - val_Recall: 0.8288 - val_loss: 0.0559
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.7889 - loss: 0.0593 - val_Recall: 0.8423 - val_loss: 0.0552
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8107 - loss: 0.0586 - val_Recall: 0.8333 - val_loss: 0.0569
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.8211 - loss: 0.0512 - val_Recall: 0.8423 - val_loss: 0.0546
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.8335 - loss: 0.0505 - val_Recall: 0.8333 - val_loss: 0.0543
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.8234 - loss: 0.0527 - val_Recall: 0.8378 - val_loss: 0.0541
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8519 - loss: 0.0480 - val_Recall: 0.8378 - val_loss: 0.0553
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.8390 - loss: 0.0461 - val_Recall: 0.8423 - val_loss: 0.0532
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8504 - loss: 0.0483 - val_Recall: 0.8468 - val_loss: 0.0530
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.8430 - loss: 0.0506 - val_Recall: 0.8378 - val_loss: 0.0550
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8305 - loss: 0.0530 - val_Recall: 0.8423 - val_loss: 0.0534
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8328 - loss: 0.0497 - val_Recall: 0.8468 - val_loss: 0.0540
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8599 - loss: 0.0416 - val_Recall: 0.8423 - val_loss: 0.0533
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 327us/step - Recall: 0.8539 - loss: 0.0427 - val_Recall: 0.8468 - val_loss: 0.0526
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8475 - loss: 0.0454 - val_Recall: 0.8423 - val_loss: 0.0535
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.8439 - loss: 0.0423 - val_Recall: 0.8468 - val_loss: 0.0531
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8474 - loss: 0.0506 - val_Recall: 0.8514 - val_loss: 0.0530
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8615 - loss: 0.0441 - val_Recall: 0.8514 - val_loss: 0.0526
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8546 - loss: 0.0460 - val_Recall: 0.8423 - val_loss: 0.0533
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.8564 - loss: 0.0473 - val_Recall: 0.8468 - val_loss: 0.0547
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.8672 - loss: 0.0455 - val_Recall: 0.8468 - val_loss: 0.0532
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8649 - loss: 0.0430 - val_Recall: 0.8514 - val_loss: 0.0526
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8842 - loss: 0.0382 - val_Recall: 0.8604 - val_loss: 0.0522
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8665 - loss: 0.0454 - val_Recall: 0.8604 - val_loss: 0.0523
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.8820 - loss: 0.0380 - val_Recall: 0.8604 - val_loss: 0.0528
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 325us/step - Recall: 0.8672 - loss: 0.0425 - val_Recall: 0.8649 - val_loss: 0.0522
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8836 - loss: 0.0377 - val_Recall: 0.8604 - val_loss: 0.0523
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8790 - loss: 0.0404 - val_Recall: 0.8694 - val_loss: 0.0523
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8648 - loss: 0.0446 - val_Recall: 0.8739 - val_loss: 0.0528
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8733 - loss: 0.0417 - val_Recall: 0.8739 - val_loss: 0.0520
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.8861 - loss: 0.0365 - val_Recall: 0.8649 - val_loss: 0.0525
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8588 - loss: 0.0413 - val_Recall: 0.8739 - val_loss: 0.0531
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8818 - loss: 0.0387 - val_Recall: 0.8649 - val_loss: 0.0534
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8714 - loss: 0.0406 - val_Recall: 0.8649 - val_loss: 0.0532
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8665 - loss: 0.0455 - val_Recall: 0.8649 - val_loss: 0.0536
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.8700 - loss: 0.0408 - val_Recall: 0.8649 - val_loss: 0.0552
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.8816 - loss: 0.0366 - val_Recall: 0.8649 - val_loss: 0.0550
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 324us/step - Recall: 0.8851 - loss: 0.0349 - val_Recall: 0.8694 - val_loss: 0.0537

print("Time taken in seconds ",end-start)

Time taken in seconds  8.56450891494751

plot(history,'loss')

plot(history,'Recall')

model_3_train_perf = model_performance_classification(model_3, x_train, y_train) #threshold = 0.5
print("Model 3 Train Performance")
print(model_3_train_perf)
print("-"*50)
print("Model 3 Validation Performance")
model_3_val_perf = model_performance_classification(model_3,x_val,y_val)
print(model_3_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 144us/step
Model 3 Train Performance
   Accuracy    Recall  Precision  F1 Score
0   0.99375  0.948463   0.991422  0.968834
--------------------------------------------------
Model 3 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 166us/step
   Accuracy    Recall  Precision  F1 Score
0    0.9915  0.934023    0.98356  0.957281

#let's get classification report on y train and y val for Model 3
y_train_pred_3 = model_3.predict(x_train)
y_val_pred_3 = model_3.predict(x_val)
print("Classification Report - Train data Model_3",end="\n\n")
cr_train_model_3 = classification_report(y_train,y_train_pred_3>0.5)
print(cr_train_model_3)
print("Classification Report - Validation data Model_3",end="\n\n")
cr_val_model_3 = classification_report(y_val,y_val_pred_3>0.5)
print(cr_val_model_3)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 151us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 173us/step
Classification Report - Train data Model_3

              precision    recall  f1-score   support

           0       0.99      1.00      1.00     15112
           1       0.99      0.90      0.94       888

    accuracy                           0.99     16000
   macro avg       0.99      0.95      0.97     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_3

              precision    recall  f1-score   support

           0       0.99      1.00      1.00      3778
           1       0.97      0.87      0.92       222

    accuracy                           0.99      4000
   macro avg       0.98      0.93      0.96      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_3.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 176us/step

results.loc[3] = [3,[128,64,64],["relu","relu","relu"],50,32,"sgd",[0.01, "-"],"-","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

# Calculate class weights for imbalanced dataset
cw = (y_train.shape[0]) / np.bincount(y_train.astype(int)) # Convert y_train to integers

# Create a dictionary mapping class indices to their respective class weights
cw_dict = {}
for i in range(cw.shape[0]):
    cw_dict[i] = cw[i]

cw_dict

{0: 1.0587612493382743, 1: 18.01801801801802}

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_4 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_4.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_4.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_4.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_4.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_4.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_4.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#SGD Optiomizer
optimizer = keras.optimizers.SGD()    # defining SGD as the optimizer to be used
model_4.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_4.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 439us/step - Recall: 0.8212 - loss: 0.8307 - val_Recall: 0.8874 - val_loss: 0.2598
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.8902 - loss: 0.5208 - val_Recall: 0.8874 - val_loss: 0.2392
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9039 - loss: 0.4192 - val_Recall: 0.8829 - val_loss: 0.2027
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9008 - loss: 0.4242 - val_Recall: 0.8874 - val_loss: 0.1191
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9017 - loss: 0.4073 - val_Recall: 0.9009 - val_loss: 0.1487
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9115 - loss: 0.3735 - val_Recall: 0.8964 - val_loss: 0.1968
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.8996 - loss: 0.3732 - val_Recall: 0.8874 - val_loss: 0.1157
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9065 - loss: 0.3482 - val_Recall: 0.8919 - val_loss: 0.1879
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9118 - loss: 0.3780 - val_Recall: 0.9009 - val_loss: 0.1402
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9205 - loss: 0.3428 - val_Recall: 0.8919 - val_loss: 0.1305
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.8920 - loss: 0.3725 - val_Recall: 0.8919 - val_loss: 0.1062
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 317us/step - Recall: 0.9045 - loss: 0.3278 - val_Recall: 0.8919 - val_loss: 0.1247
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9121 - loss: 0.3249 - val_Recall: 0.8919 - val_loss: 0.1439
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 323us/step - Recall: 0.9066 - loss: 0.3283 - val_Recall: 0.8874 - val_loss: 0.1218
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9074 - loss: 0.3492 - val_Recall: 0.8964 - val_loss: 0.1152
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9108 - loss: 0.3124 - val_Recall: 0.8919 - val_loss: 0.1065
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9009 - loss: 0.3493 - val_Recall: 0.8919 - val_loss: 0.1138
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9171 - loss: 0.3037 - val_Recall: 0.9009 - val_loss: 0.1623
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9143 - loss: 0.3139 - val_Recall: 0.8964 - val_loss: 0.1379
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9253 - loss: 0.2829 - val_Recall: 0.8919 - val_loss: 0.1088
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9148 - loss: 0.3155 - val_Recall: 0.9054 - val_loss: 0.1530
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9166 - loss: 0.2788 - val_Recall: 0.8964 - val_loss: 0.1282
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9054 - loss: 0.3134 - val_Recall: 0.8874 - val_loss: 0.1089
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9192 - loss: 0.2828 - val_Recall: 0.8919 - val_loss: 0.1232
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9052 - loss: 0.3074 - val_Recall: 0.8919 - val_loss: 0.1094
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9252 - loss: 0.2696 - val_Recall: 0.8919 - val_loss: 0.1328
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9204 - loss: 0.2734 - val_Recall: 0.9009 - val_loss: 0.1442
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9229 - loss: 0.2718 - val_Recall: 0.8829 - val_loss: 0.1051
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9238 - loss: 0.2598 - val_Recall: 0.8964 - val_loss: 0.1007
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9232 - loss: 0.2869 - val_Recall: 0.9009 - val_loss: 0.1579
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9083 - loss: 0.2964 - val_Recall: 0.8919 - val_loss: 0.1282
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9205 - loss: 0.2689 - val_Recall: 0.8874 - val_loss: 0.1028
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9101 - loss: 0.2843 - val_Recall: 0.9009 - val_loss: 0.1693
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 322us/step - Recall: 0.9151 - loss: 0.3001 - val_Recall: 0.8874 - val_loss: 0.1153
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9300 - loss: 0.2463 - val_Recall: 0.8919 - val_loss: 0.1109
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9223 - loss: 0.2401 - val_Recall: 0.8919 - val_loss: 0.1203
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9279 - loss: 0.2393 - val_Recall: 0.8964 - val_loss: 0.1216
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9164 - loss: 0.2636 - val_Recall: 0.8964 - val_loss: 0.1389
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9225 - loss: 0.2582 - val_Recall: 0.8919 - val_loss: 0.1043
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9197 - loss: 0.2494 - val_Recall: 0.8919 - val_loss: 0.1003
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 321us/step - Recall: 0.9254 - loss: 0.2498 - val_Recall: 0.8964 - val_loss: 0.1129
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9088 - loss: 0.2779 - val_Recall: 0.8919 - val_loss: 0.1320
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9282 - loss: 0.2360 - val_Recall: 0.8919 - val_loss: 0.1024
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 318us/step - Recall: 0.9184 - loss: 0.2546 - val_Recall: 0.8919 - val_loss: 0.0963
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9142 - loss: 0.2614 - val_Recall: 0.8964 - val_loss: 0.1444
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9372 - loss: 0.2167 - val_Recall: 0.8964 - val_loss: 0.1056
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9283 - loss: 0.2475 - val_Recall: 0.8964 - val_loss: 0.1172
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 319us/step - Recall: 0.9184 - loss: 0.2520 - val_Recall: 0.8964 - val_loss: 0.1421
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9247 - loss: 0.2428 - val_Recall: 0.8964 - val_loss: 0.1270
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 320us/step - Recall: 0.9397 - loss: 0.2074 - val_Recall: 0.8919 - val_loss: 0.1085

print("Time taken in seconds ",end-start)

Time taken in seconds  8.464919090270996

plot(history,'loss')

plot(history,'Recall')

model_4_train_perf = model_performance_classification(model_4, x_train, y_train) #threshold = 0.5
print("Model 4 Train Performance")
print(model_4_train_perf)
print("-"*50)
print("Model 4 Validation Performance")
model_4_val_perf = model_performance_classification(model_4,x_val,y_val)
print(model_4_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 147us/step
Model 4 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.989812  0.961218   0.943856  0.952347
--------------------------------------------------
Model 4 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 162us/step
   Accuracy    Recall  Precision  F1 Score
0    0.9875  0.942505   0.938786  0.940637

#let's get classification report on y train and y val for Model 4
y_train_pred_4 = model_4.predict(x_train)
y_val_pred_4 = model_4.predict(x_val)
print("Classification Report - Train data Model_4",end="\n\n")
cr_train_model_4 = classification_report(y_train,y_train_pred_4>0.5)
print(cr_train_model_4)
print("Classification Report - Validation data Model_4",end="\n\n")
cr_val_model_4 = classification_report(y_val,y_val_pred_4>0.5)
print(cr_val_model_4)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 151us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 176us/step
Classification Report - Train data Model_4

              precision    recall  f1-score   support

           0       1.00      0.99      0.99     15112
           1       0.89      0.93      0.91       888

    accuracy                           0.99     16000
   macro avg       0.94      0.96      0.95     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_4

              precision    recall  f1-score   support

           0       0.99      0.99      0.99      3778
           1       0.88      0.89      0.89       222

    accuracy                           0.99      4000
   macro avg       0.94      0.94      0.94      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_4.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 179us/step

results.loc[4] = [3,[128,64,64],["relu","relu","relu"],50,32,"sgd",[0.01, "-"],"Xavier - CW ","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_5 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_5.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_5.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_5.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_5.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_5.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_5.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#SGD Optiomizer
mom = 0.3
optimizer = keras.optimizers.SGD(momentum=mom)    # defining SGD as the optimizer to be used
model_5.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_5.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 457us/step - Recall: 0.8293 - loss: 0.8449 - val_Recall: 0.8874 - val_loss: 0.2408
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 331us/step - Recall: 0.8989 - loss: 0.4688 - val_Recall: 0.8874 - val_loss: 0.2295
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.8926 - loss: 0.4525 - val_Recall: 0.8829 - val_loss: 0.1341
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 330us/step - Recall: 0.8897 - loss: 0.4305 - val_Recall: 0.9009 - val_loss: 0.1757
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9030 - loss: 0.4031 - val_Recall: 0.8829 - val_loss: 0.1624
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 331us/step - Recall: 0.9040 - loss: 0.3513 - val_Recall: 0.9009 - val_loss: 0.1511
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9079 - loss: 0.3484 - val_Recall: 0.8829 - val_loss: 0.1854
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.8962 - loss: 0.3684 - val_Recall: 0.8784 - val_loss: 0.1131
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.9214 - loss: 0.3132 - val_Recall: 0.8964 - val_loss: 0.1699
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 332us/step - Recall: 0.9108 - loss: 0.3461 - val_Recall: 0.8829 - val_loss: 0.1292
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.9150 - loss: 0.3250 - val_Recall: 0.8784 - val_loss: 0.1670
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 330us/step - Recall: 0.9219 - loss: 0.3285 - val_Recall: 0.8829 - val_loss: 0.1087
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 332us/step - Recall: 0.9143 - loss: 0.3222 - val_Recall: 0.8829 - val_loss: 0.1094
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 332us/step - Recall: 0.9014 - loss: 0.3308 - val_Recall: 0.8964 - val_loss: 0.1953
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 332us/step - Recall: 0.9206 - loss: 0.2950 - val_Recall: 0.8829 - val_loss: 0.1426
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 383us/step - Recall: 0.9203 - loss: 0.2973 - val_Recall: 0.8784 - val_loss: 0.1159
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 336us/step - Recall: 0.9115 - loss: 0.3155 - val_Recall: 0.8874 - val_loss: 0.1341
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9134 - loss: 0.3245 - val_Recall: 0.8829 - val_loss: 0.1180
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9209 - loss: 0.2860 - val_Recall: 0.8964 - val_loss: 0.1536
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.8997 - loss: 0.3283 - val_Recall: 0.8964 - val_loss: 0.1518
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9292 - loss: 0.2760 - val_Recall: 0.9009 - val_loss: 0.1443
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9120 - loss: 0.2945 - val_Recall: 0.8964 - val_loss: 0.1311
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9244 - loss: 0.2810 - val_Recall: 0.8874 - val_loss: 0.1345
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.9235 - loss: 0.2720 - val_Recall: 0.8874 - val_loss: 0.1137
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9248 - loss: 0.2619 - val_Recall: 0.8874 - val_loss: 0.1171
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9047 - loss: 0.3394 - val_Recall: 0.8829 - val_loss: 0.0974
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9174 - loss: 0.3080 - val_Recall: 0.8919 - val_loss: 0.1254
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9265 - loss: 0.2291 - val_Recall: 0.8874 - val_loss: 0.1280
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9206 - loss: 0.2962 - val_Recall: 0.9009 - val_loss: 0.1720
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9164 - loss: 0.2872 - val_Recall: 0.8829 - val_loss: 0.1204
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9116 - loss: 0.2929 - val_Recall: 0.8874 - val_loss: 0.1348
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9229 - loss: 0.2596 - val_Recall: 0.8829 - val_loss: 0.0931
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9050 - loss: 0.3151 - val_Recall: 0.8964 - val_loss: 0.1739
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 337us/step - Recall: 0.9165 - loss: 0.2935 - val_Recall: 0.9009 - val_loss: 0.1323
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.8965 - loss: 0.3071 - val_Recall: 0.8874 - val_loss: 0.1246
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9258 - loss: 0.2533 - val_Recall: 0.8919 - val_loss: 0.1370
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9064 - loss: 0.2946 - val_Recall: 0.8964 - val_loss: 0.1178
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9246 - loss: 0.2566 - val_Recall: 0.8964 - val_loss: 0.1050
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9358 - loss: 0.2470 - val_Recall: 0.8919 - val_loss: 0.1314
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9171 - loss: 0.2983 - val_Recall: 0.8964 - val_loss: 0.0950
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9225 - loss: 0.2524 - val_Recall: 0.8919 - val_loss: 0.1375
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9183 - loss: 0.2804 - val_Recall: 0.8874 - val_loss: 0.0967
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9247 - loss: 0.2596 - val_Recall: 0.8874 - val_loss: 0.1181
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 334us/step - Recall: 0.9189 - loss: 0.2602 - val_Recall: 0.8964 - val_loss: 0.1239
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9280 - loss: 0.2422 - val_Recall: 0.8919 - val_loss: 0.1009
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9253 - loss: 0.2393 - val_Recall: 0.8964 - val_loss: 0.1437
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 333us/step - Recall: 0.9118 - loss: 0.2616 - val_Recall: 0.9009 - val_loss: 0.1289
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9249 - loss: 0.2536 - val_Recall: 0.9009 - val_loss: 0.1350
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9279 - loss: 0.2413 - val_Recall: 0.8964 - val_loss: 0.1693
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 335us/step - Recall: 0.9356 - loss: 0.2391 - val_Recall: 0.8874 - val_loss: 0.1120

print("Time taken in seconds ",end-start)

Time taken in seconds  8.886943817138672

plot(history,'loss')

plot(history,'Recall')

model_5_train_perf = model_performance_classification(model_5, x_train, y_train) #threshold = 0.5
print("Model 5 Train Performance")
print(model_5_train_perf)
print("-"*50)
print("Model 5 Validation Performance")
model_5_val_perf = model_performance_classification(model_5,x_val,y_val)
print(model_5_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 144us/step
Model 5 Train Performance
   Accuracy    Recall  Precision  F1 Score
0    0.9915  0.962642   0.956836  0.959718
--------------------------------------------------
Model 5 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 164us/step
   Accuracy    Recall  Precision  F1 Score
0   0.98825  0.940782   0.946466  0.943603

#let's get classification report on y train and y val for Model 5
y_train_pred_5 = model_5.predict(x_train)
y_val_pred_5 = model_5.predict(x_val)
print("Classification Report - Train data Model_5",end="\n\n")
cr_train_model_5 = classification_report(y_train,y_train_pred_5>0.5)
print(cr_train_model_5)
print("Classification Report - Validation data Model_5",end="\n\n")
cr_val_model_5 = classification_report(y_val,y_val_pred_5>0.5)
print(cr_val_model_5)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 150us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 172us/step
Classification Report - Train data Model_5

              precision    recall  f1-score   support

           0       1.00      1.00      1.00     15112
           1       0.92      0.93      0.92       888

    accuracy                           0.99     16000
   macro avg       0.96      0.96      0.96     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_5

              precision    recall  f1-score   support

           0       0.99      0.99      0.99      3778
           1       0.90      0.89      0.89       222

    accuracy                           0.99      4000
   macro avg       0.95      0.94      0.94      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_5.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 174us/step

results.loc[5] = [3,[128,64,64],["relu","relu","relu"],50,32,"sgd",[0.01, "0.3"],"Xavier - CW ","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_6 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_6.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_6.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_6.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_6.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_6.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_6.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#Adam Optiomizer 
lr = 0.001
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_6.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_6.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 501us/step - Recall: 0.7646 - loss: 0.8452 - val_Recall: 0.8784 - val_loss: 0.1619
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8737 - loss: 0.5518 - val_Recall: 0.8919 - val_loss: 0.1525
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9103 - loss: 0.4068 - val_Recall: 0.8784 - val_loss: 0.1448
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 366us/step - Recall: 0.8930 - loss: 0.4071 - val_Recall: 0.8829 - val_loss: 0.1281
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 365us/step - Recall: 0.9028 - loss: 0.3705 - val_Recall: 0.8874 - val_loss: 0.1164
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 366us/step - Recall: 0.9094 - loss: 0.3289 - val_Recall: 0.8829 - val_loss: 0.1183
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9139 - loss: 0.3585 - val_Recall: 0.8874 - val_loss: 0.1834
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 369us/step - Recall: 0.9016 - loss: 0.3636 - val_Recall: 0.8874 - val_loss: 0.0956
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 369us/step - Recall: 0.9052 - loss: 0.3671 - val_Recall: 0.9099 - val_loss: 0.1792
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9219 - loss: 0.3126 - val_Recall: 0.9054 - val_loss: 0.1824
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9142 - loss: 0.3117 - val_Recall: 0.9099 - val_loss: 0.1883
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9020 - loss: 0.3718 - val_Recall: 0.8694 - val_loss: 0.0791
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9131 - loss: 0.3266 - val_Recall: 0.8874 - val_loss: 0.1047
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9176 - loss: 0.2983 - val_Recall: 0.8874 - val_loss: 0.1280
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9060 - loss: 0.3272 - val_Recall: 0.8919 - val_loss: 0.1352
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9015 - loss: 0.3389 - val_Recall: 0.8919 - val_loss: 0.1153
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9174 - loss: 0.3103 - val_Recall: 0.8874 - val_loss: 0.1582
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9102 - loss: 0.2932 - val_Recall: 0.8784 - val_loss: 0.1577
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9103 - loss: 0.3013 - val_Recall: 0.8874 - val_loss: 0.1351
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9207 - loss: 0.2991 - val_Recall: 0.8964 - val_loss: 0.1587
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 369us/step - Recall: 0.9155 - loss: 0.2699 - val_Recall: 0.8964 - val_loss: 0.1644
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9211 - loss: 0.2969 - val_Recall: 0.9009 - val_loss: 0.1517
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9195 - loss: 0.2960 - val_Recall: 0.8829 - val_loss: 0.0952
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9021 - loss: 0.3202 - val_Recall: 0.8964 - val_loss: 0.1688
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9312 - loss: 0.2878 - val_Recall: 0.8874 - val_loss: 0.1633
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9208 - loss: 0.2565 - val_Recall: 0.8829 - val_loss: 0.1511
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9142 - loss: 0.2922 - val_Recall: 0.8964 - val_loss: 0.1589
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9246 - loss: 0.2742 - val_Recall: 0.8874 - val_loss: 0.1098
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9109 - loss: 0.2806 - val_Recall: 0.8964 - val_loss: 0.1348
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9227 - loss: 0.2578 - val_Recall: 0.8874 - val_loss: 0.1126
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9291 - loss: 0.2629 - val_Recall: 0.8964 - val_loss: 0.1435
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 422us/step - Recall: 0.9230 - loss: 0.2512 - val_Recall: 0.8919 - val_loss: 0.1644
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9298 - loss: 0.2407 - val_Recall: 0.8784 - val_loss: 0.1320
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9134 - loss: 0.2795 - val_Recall: 0.9009 - val_loss: 0.1485
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9311 - loss: 0.2173 - val_Recall: 0.8919 - val_loss: 0.1494
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9317 - loss: 0.2248 - val_Recall: 0.9009 - val_loss: 0.1513
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9185 - loss: 0.2493 - val_Recall: 0.8919 - val_loss: 0.1142
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9111 - loss: 0.2606 - val_Recall: 0.8874 - val_loss: 0.1717
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9423 - loss: 0.2089 - val_Recall: 0.9009 - val_loss: 0.2404
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9326 - loss: 0.2409 - val_Recall: 0.8784 - val_loss: 0.1390
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9258 - loss: 0.2303 - val_Recall: 0.8874 - val_loss: 0.1295
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 370us/step - Recall: 0.9176 - loss: 0.2367 - val_Recall: 0.8829 - val_loss: 0.1028
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9233 - loss: 0.2285 - val_Recall: 0.8874 - val_loss: 0.1509
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9160 - loss: 0.2343 - val_Recall: 0.8919 - val_loss: 0.1429
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9324 - loss: 0.2301 - val_Recall: 0.9009 - val_loss: 0.1330
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9304 - loss: 0.2121 - val_Recall: 0.8874 - val_loss: 0.1377
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9294 - loss: 0.2247 - val_Recall: 0.8784 - val_loss: 0.1667
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9309 - loss: 0.2240 - val_Recall: 0.8784 - val_loss: 0.1366
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9225 - loss: 0.2176 - val_Recall: 0.8784 - val_loss: 0.1274
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9259 - loss: 0.2271 - val_Recall: 0.9009 - val_loss: 0.1431

print("Time taken in seconds ",end-start)

Time taken in seconds  9.925749063491821

plot(history,'loss')

plot(history,'Recall')

model_6_train_perf = model_performance_classification(model_6, x_train, y_train) #threshold = 0.5
print("Model 6 Train Performance")
print(model_6_train_perf)
print("-"*50)
print("Model 6 Validation Performance")
model_6_val_perf = model_performance_classification(model_6,x_val,y_val)
print(model_6_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 149us/step
Model 6 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.987125  0.966685   0.919863  0.941851
--------------------------------------------------
Model 6 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 170us/step
   Accuracy    Recall  Precision  F1 Score
0    0.9825  0.944098   0.900294  0.920884

#let's get classification report on y train and y val for Model 5
y_train_pred_6 = model_6.predict(x_train)
y_val_pred_6 = model_6.predict(x_val)
print("Classification Report - Train data Model_6",end="\n\n")
cr_train_model_6 = classification_report(y_train,y_train_pred_6>0.5)
print(cr_train_model_6)
print("Classification Report - Validation data Model_6",end="\n\n")
cr_val_model_6 = classification_report(y_val,y_val_pred_6>0.5)
print(cr_val_model_6)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 153us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 171us/step
Classification Report - Train data Model_6

              precision    recall  f1-score   support

           0       1.00      0.99      0.99     15112
           1       0.84      0.94      0.89       888

    accuracy                           0.99     16000
   macro avg       0.92      0.97      0.94     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_6

              precision    recall  f1-score   support

           0       0.99      0.99      0.99      3778
           1       0.81      0.90      0.85       222

    accuracy                           0.98      4000
   macro avg       0.90      0.94      0.92      4000
weighted avg       0.98      0.98      0.98      4000

# Predict probabilities using confusion matrix 
y_pred = model_6.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 169us/step

results.loc[6] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[0.001, "-"],"Xavier - CW ","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_7 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_7.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_7.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_7.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_7.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_7.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_7.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#Adam Optiomizer 
lr = 1e-4
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_7.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_7.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 498us/step - Recall: 0.3989 - loss: 1.6584 - val_Recall: 0.8784 - val_loss: 0.3903
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 379us/step - Recall: 0.8562 - loss: 0.7610 - val_Recall: 0.8919 - val_loss: 0.2993
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 382us/step - Recall: 0.8832 - loss: 0.6718 - val_Recall: 0.8964 - val_loss: 0.2529
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.8999 - loss: 0.5691 - val_Recall: 0.8874 - val_loss: 0.1933
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8809 - loss: 0.5674 - val_Recall: 0.8919 - val_loss: 0.1943
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9027 - loss: 0.4760 - val_Recall: 0.8919 - val_loss: 0.1733
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8943 - loss: 0.5174 - val_Recall: 0.8919 - val_loss: 0.1634
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8815 - loss: 0.4944 - val_Recall: 0.8919 - val_loss: 0.1670
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.9031 - loss: 0.4581 - val_Recall: 0.8919 - val_loss: 0.1480
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 371us/step - Recall: 0.8910 - loss: 0.4551 - val_Recall: 0.8919 - val_loss: 0.1491
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9010 - loss: 0.4577 - val_Recall: 0.8874 - val_loss: 0.1221
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9158 - loss: 0.4094 - val_Recall: 0.8919 - val_loss: 0.1534
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9038 - loss: 0.3917 - val_Recall: 0.8919 - val_loss: 0.1349
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8975 - loss: 0.4176 - val_Recall: 0.8964 - val_loss: 0.1551
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.8947 - loss: 0.4009 - val_Recall: 0.8919 - val_loss: 0.1415
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9008 - loss: 0.3887 - val_Recall: 0.8874 - val_loss: 0.1311
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 421us/step - Recall: 0.9086 - loss: 0.3637 - val_Recall: 0.8829 - val_loss: 0.1178
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9045 - loss: 0.3845 - val_Recall: 0.8919 - val_loss: 0.1337
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9064 - loss: 0.3667 - val_Recall: 0.8874 - val_loss: 0.1232
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9162 - loss: 0.3517 - val_Recall: 0.8874 - val_loss: 0.1274
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9079 - loss: 0.3685 - val_Recall: 0.8874 - val_loss: 0.1370
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8977 - loss: 0.3802 - val_Recall: 0.8874 - val_loss: 0.1040
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8822 - loss: 0.3937 - val_Recall: 0.8874 - val_loss: 0.1233
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9220 - loss: 0.3266 - val_Recall: 0.8874 - val_loss: 0.1571
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9001 - loss: 0.3871 - val_Recall: 0.8874 - val_loss: 0.1206
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9272 - loss: 0.3059 - val_Recall: 0.8874 - val_loss: 0.1199
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9263 - loss: 0.3016 - val_Recall: 0.8919 - val_loss: 0.1550
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9185 - loss: 0.3121 - val_Recall: 0.8874 - val_loss: 0.1295
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.9136 - loss: 0.3377 - val_Recall: 0.8874 - val_loss: 0.1168
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.9109 - loss: 0.3414 - val_Recall: 0.8874 - val_loss: 0.1213
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9189 - loss: 0.2864 - val_Recall: 0.8874 - val_loss: 0.1491
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9274 - loss: 0.3012 - val_Recall: 0.8874 - val_loss: 0.1163
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9213 - loss: 0.3107 - val_Recall: 0.8874 - val_loss: 0.1231
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9168 - loss: 0.3201 - val_Recall: 0.8874 - val_loss: 0.1116
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9244 - loss: 0.3018 - val_Recall: 0.8874 - val_loss: 0.1135
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9068 - loss: 0.3124 - val_Recall: 0.8874 - val_loss: 0.1249
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9273 - loss: 0.2753 - val_Recall: 0.8874 - val_loss: 0.1270
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9082 - loss: 0.3276 - val_Recall: 0.8874 - val_loss: 0.1137
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9145 - loss: 0.3000 - val_Recall: 0.8874 - val_loss: 0.1052
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9066 - loss: 0.3058 - val_Recall: 0.8919 - val_loss: 0.1441
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9301 - loss: 0.2712 - val_Recall: 0.8919 - val_loss: 0.1347
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.9132 - loss: 0.2769 - val_Recall: 0.8874 - val_loss: 0.1232
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9145 - loss: 0.2735 - val_Recall: 0.8874 - val_loss: 0.1431
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 373us/step - Recall: 0.9136 - loss: 0.3110 - val_Recall: 0.8874 - val_loss: 0.0922
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9069 - loss: 0.3045 - val_Recall: 0.8874 - val_loss: 0.1270
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9184 - loss: 0.2783 - val_Recall: 0.8874 - val_loss: 0.1225
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.9277 - loss: 0.2828 - val_Recall: 0.8874 - val_loss: 0.1148
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9272 - loss: 0.2515 - val_Recall: 0.8874 - val_loss: 0.1208
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 372us/step - Recall: 0.9313 - loss: 0.2552 - val_Recall: 0.8874 - val_loss: 0.1319
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.9165 - loss: 0.2655 - val_Recall: 0.8874 - val_loss: 0.1101

print("Time taken in seconds ",end-start)

Time taken in seconds  10.007179021835327

plot(history,'loss')

plot(history,'Recall')

model_7_train_perf = model_performance_classification(model_7, x_train, y_train) #threshold = 0.5
print("Model 7 Train Performance")
print(model_7_train_perf)
print("-"*50)
print("Model 7 Validation Performance")
model_7_val_perf = model_performance_classification(model_7,x_val,y_val)
print(model_7_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 145us/step
Model 7 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.989313  0.958304   0.941949  0.949956
--------------------------------------------------
Model 7 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 167us/step
   Accuracy    Recall  Precision  F1 Score
0   0.98725  0.940253   0.938395  0.939321

#let's get classification report on y train and y val for Model 7
y_train_pred_7 = model_7.predict(x_train)
y_val_pred_7 = model_7.predict(x_val)
print("Classification Report - Train data Model_7",end="\n\n")
cr_train_model_7 = classification_report(y_train,y_train_pred_7>0.5)
print(cr_train_model_7)
print("Classification Report - Validation data Model_7",end="\n\n")
cr_val_model_7 = classification_report(y_val,y_val_pred_7>0.5)
print(cr_val_model_7)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 148us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 174us/step
Classification Report - Train data Model_7

              precision    recall  f1-score   support

           0       1.00      0.99      0.99     15112
           1       0.89      0.92      0.91       888

    accuracy                           0.99     16000
   macro avg       0.94      0.96      0.95     16000
weighted avg       0.99      0.99      0.99     16000

Classification Report - Validation data Model_7

              precision    recall  f1-score   support

           0       0.99      0.99      0.99      3778
           1       0.88      0.89      0.89       222

    accuracy                           0.99      4000
   macro avg       0.94      0.94      0.94      4000
weighted avg       0.99      0.99      0.99      4000

# Predict probabilities using confusion matrix 
y_pred = model_7.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 168us/step

results.loc[7] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[1e-04, "-"],"Xavier - CW ","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_8 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_8.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_8.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_8.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_8.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_8.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_8.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 17,729 (69.25 KB)

 Trainable params: 17,729 (69.25 KB)

 Non-trainable params: 0 (0.00 B)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_8.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_8.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 499us/step - Recall: 0.8998 - loss: 1.6624 - val_Recall: 0.7613 - val_loss: 0.6448
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.7767 - loss: 1.2090 - val_Recall: 0.7523 - val_loss: 0.4872
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.7361 - loss: 1.1025 - val_Recall: 0.7658 - val_loss: 0.4323
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.7975 - loss: 0.9568 - val_Recall: 0.7793 - val_loss: 0.4050
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8031 - loss: 0.9381 - val_Recall: 0.7928 - val_loss: 0.3812
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.7670 - loss: 0.9312 - val_Recall: 0.8153 - val_loss: 0.3619
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8061 - loss: 0.8524 - val_Recall: 0.8198 - val_loss: 0.3483
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.8294 - loss: 0.8025 - val_Recall: 0.8243 - val_loss: 0.3344
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8443 - loss: 0.8099 - val_Recall: 0.8423 - val_loss: 0.3213
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8341 - loss: 0.7757 - val_Recall: 0.8468 - val_loss: 0.3100
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8546 - loss: 0.8072 - val_Recall: 0.8468 - val_loss: 0.2928
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8867 - loss: 0.6780 - val_Recall: 0.8649 - val_loss: 0.2864
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8616 - loss: 0.6897 - val_Recall: 0.8649 - val_loss: 0.2751
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8588 - loss: 0.7310 - val_Recall: 0.8694 - val_loss: 0.2670
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8832 - loss: 0.6706 - val_Recall: 0.8784 - val_loss: 0.2614
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 374us/step - Recall: 0.8702 - loss: 0.6993 - val_Recall: 0.8829 - val_loss: 0.2510
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8840 - loss: 0.6585 - val_Recall: 0.8829 - val_loss: 0.2490
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.8713 - loss: 0.6453 - val_Recall: 0.8874 - val_loss: 0.2429
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 381us/step - Recall: 0.8958 - loss: 0.6416 - val_Recall: 0.8829 - val_loss: 0.2392
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8577 - loss: 0.6586 - val_Recall: 0.8829 - val_loss: 0.2319
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8806 - loss: 0.5998 - val_Recall: 0.8829 - val_loss: 0.2281
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8830 - loss: 0.5773 - val_Recall: 0.8829 - val_loss: 0.2254
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8818 - loss: 0.6106 - val_Recall: 0.8829 - val_loss: 0.2200
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.8698 - loss: 0.6071 - val_Recall: 0.8829 - val_loss: 0.2156
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.8865 - loss: 0.5822 - val_Recall: 0.8829 - val_loss: 0.2102
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8685 - loss: 0.5977 - val_Recall: 0.8829 - val_loss: 0.2074
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8977 - loss: 0.5267 - val_Recall: 0.8874 - val_loss: 0.2054
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 428us/step - Recall: 0.8972 - loss: 0.5788 - val_Recall: 0.8874 - val_loss: 0.1997
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 380us/step - Recall: 0.8907 - loss: 0.5531 - val_Recall: 0.8874 - val_loss: 0.1963
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8763 - loss: 0.5732 - val_Recall: 0.8874 - val_loss: 0.1919
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8908 - loss: 0.5556 - val_Recall: 0.8874 - val_loss: 0.1914
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.9036 - loss: 0.5317 - val_Recall: 0.8874 - val_loss: 0.1910
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.9013 - loss: 0.5161 - val_Recall: 0.8874 - val_loss: 0.1893
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8837 - loss: 0.5157 - val_Recall: 0.8874 - val_loss: 0.1839
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8720 - loss: 0.5710 - val_Recall: 0.8874 - val_loss: 0.1829
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8659 - loss: 0.5846 - val_Recall: 0.8919 - val_loss: 0.1782
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 379us/step - Recall: 0.8847 - loss: 0.5319 - val_Recall: 0.8919 - val_loss: 0.1785
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 380us/step - Recall: 0.8958 - loss: 0.4959 - val_Recall: 0.8919 - val_loss: 0.1768
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 379us/step - Recall: 0.8928 - loss: 0.5311 - val_Recall: 0.8919 - val_loss: 0.1722
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.9001 - loss: 0.4968 - val_Recall: 0.8919 - val_loss: 0.1718
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8988 - loss: 0.5197 - val_Recall: 0.8919 - val_loss: 0.1702
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8733 - loss: 0.5366 - val_Recall: 0.8919 - val_loss: 0.1702
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 381us/step - Recall: 0.8837 - loss: 0.5061 - val_Recall: 0.8919 - val_loss: 0.1693
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8732 - loss: 0.5262 - val_Recall: 0.8919 - val_loss: 0.1694
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8954 - loss: 0.5390 - val_Recall: 0.8919 - val_loss: 0.1656
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.8954 - loss: 0.4729 - val_Recall: 0.8919 - val_loss: 0.1668
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 377us/step - Recall: 0.9042 - loss: 0.4597 - val_Recall: 0.8919 - val_loss: 0.1643
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 378us/step - Recall: 0.9104 - loss: 0.4624 - val_Recall: 0.8919 - val_loss: 0.1625
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 376us/step - Recall: 0.8871 - loss: 0.5189 - val_Recall: 0.8919 - val_loss: 0.1619
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 375us/step - Recall: 0.8759 - loss: 0.5361 - val_Recall: 0.8919 - val_loss: 0.1611

print("Time taken in seconds ",end-start)

Time taken in seconds  10.07526683807373

plot(history,'loss')

plot(history,'Recall')

model_8_train_perf = model_performance_classification(model_8, x_train, y_train) #threshold = 0.5
print("Model 8 Train Performance")
print(model_8_train_perf)
print("-"*50)
print("Model 8 Validation Performance")
model_8_val_perf = model_performance_classification(model_8,x_val,y_val)
print(model_8_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 144us/step
Model 8 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.971625  0.938341   0.840087  0.881904
--------------------------------------------------
Model 8 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 166us/step
   Accuracy    Recall  Precision  F1 Score
0      0.97  0.933241   0.833497  0.875704

#let's get classification report on y train and y val for Model 8
y_train_pred_8 = model_8.predict(x_train)
y_val_pred_8 = model_8.predict(x_val)
print("Classification Report - Train data Model_8",end="\n\n")
cr_train_model_8 = classification_report(y_train,y_train_pred_8>0.5)
print(cr_train_model_8)
print("Classification Report - Validation data Model_8",end="\n\n")
cr_val_model_8 = classification_report(y_val,y_val_pred_8>0.5)
print(cr_val_model_8)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 146us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 167us/step
Classification Report - Train data Model_8

              precision    recall  f1-score   support

           0       0.99      0.98      0.98     15112
           1       0.69      0.90      0.78       888

    accuracy                           0.97     16000
   macro avg       0.84      0.94      0.88     16000
weighted avg       0.98      0.97      0.97     16000

Classification Report - Validation data Model_8

              precision    recall  f1-score   support

           0       0.99      0.97      0.98      3778
           1       0.67      0.89      0.77       222

    accuracy                           0.97      4000
   macro avg       0.83      0.93      0.88      4000
weighted avg       0.98      0.97      0.97      4000

# Predict probabilities using confusion matrix 
y_pred = model_8.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 177us/step

results.loc[8] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[1e-5, "-"],"Xavier - CW ","dropout(0.3)",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_9 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_9.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1]))
model_9.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_9.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_9.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_9.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_9.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_9.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_9.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_9.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization             │ (None, 128)            │           512 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_1           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_2           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 18,753 (73.25 KB)

 Trainable params: 18,241 (71.25 KB)

 Non-trainable params: 512 (2.00 KB)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_9.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_9.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 623us/step - Recall: 0.3705 - loss: 2.4488 - val_Recall: 0.4775 - val_loss: 0.8437
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.6144 - loss: 1.7520 - val_Recall: 0.7117 - val_loss: 0.7759
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.6967 - loss: 1.4798 - val_Recall: 0.8288 - val_loss: 0.7253
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8097 - loss: 1.2481 - val_Recall: 0.8739 - val_loss: 0.6968
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8398 - loss: 1.1383 - val_Recall: 0.8829 - val_loss: 0.6541
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8642 - loss: 1.0938 - val_Recall: 0.8964 - val_loss: 0.6321
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 451us/step - Recall: 0.8417 - loss: 1.0768 - val_Recall: 0.9144 - val_loss: 0.6081
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8956 - loss: 1.0049 - val_Recall: 0.9189 - val_loss: 0.5836
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 460us/step - Recall: 0.8832 - loss: 0.9862 - val_Recall: 0.9234 - val_loss: 0.5477
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 459us/step - Recall: 0.8958 - loss: 0.9349 - val_Recall: 0.9189 - val_loss: 0.5392
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8961 - loss: 0.9025 - val_Recall: 0.9144 - val_loss: 0.5132
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.9138 - loss: 0.8744 - val_Recall: 0.9144 - val_loss: 0.4941
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.9064 - loss: 0.8623 - val_Recall: 0.9144 - val_loss: 0.4766
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.8985 - loss: 0.8444 - val_Recall: 0.9099 - val_loss: 0.4534
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9213 - loss: 0.7858 - val_Recall: 0.9099 - val_loss: 0.4383
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 461us/step - Recall: 0.9125 - loss: 0.7734 - val_Recall: 0.9144 - val_loss: 0.4266
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 457us/step - Recall: 0.8866 - loss: 0.7988 - val_Recall: 0.9144 - val_loss: 0.4069
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8878 - loss: 0.7764 - val_Recall: 0.9099 - val_loss: 0.4044
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.8927 - loss: 0.7722 - val_Recall: 0.9144 - val_loss: 0.3876
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8987 - loss: 0.7379 - val_Recall: 0.9009 - val_loss: 0.3718
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9045 - loss: 0.7177 - val_Recall: 0.9009 - val_loss: 0.3638
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8921 - loss: 0.7127 - val_Recall: 0.8964 - val_loss: 0.3453
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.8708 - loss: 0.7407 - val_Recall: 0.9009 - val_loss: 0.3495
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8928 - loss: 0.6848 - val_Recall: 0.8964 - val_loss: 0.3318
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8832 - loss: 0.6903 - val_Recall: 0.8919 - val_loss: 0.3217
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8768 - loss: 0.7081 - val_Recall: 0.8964 - val_loss: 0.3138
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8758 - loss: 0.7093 - val_Recall: 0.8964 - val_loss: 0.3057
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8977 - loss: 0.6438 - val_Recall: 0.8964 - val_loss: 0.3012
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.9095 - loss: 0.6377 - val_Recall: 0.8964 - val_loss: 0.2979
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9023 - loss: 0.6365 - val_Recall: 0.8919 - val_loss: 0.2764
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.8847 - loss: 0.6575 - val_Recall: 0.8919 - val_loss: 0.2802
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 512us/step - Recall: 0.8844 - loss: 0.6467 - val_Recall: 0.8919 - val_loss: 0.2726
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8883 - loss: 0.6207 - val_Recall: 0.8919 - val_loss: 0.2638
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.9053 - loss: 0.5844 - val_Recall: 0.8919 - val_loss: 0.2585
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.9145 - loss: 0.5755 - val_Recall: 0.8874 - val_loss: 0.2552
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8931 - loss: 0.5853 - val_Recall: 0.8919 - val_loss: 0.2567
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8817 - loss: 0.5770 - val_Recall: 0.8829 - val_loss: 0.2419
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8942 - loss: 0.6185 - val_Recall: 0.8874 - val_loss: 0.2426
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 460us/step - Recall: 0.8979 - loss: 0.5712 - val_Recall: 0.8829 - val_loss: 0.2371
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 466us/step - Recall: 0.8776 - loss: 0.6175 - val_Recall: 0.8829 - val_loss: 0.2314
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8976 - loss: 0.5780 - val_Recall: 0.8829 - val_loss: 0.2285
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8748 - loss: 0.6110 - val_Recall: 0.8829 - val_loss: 0.2231
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8651 - loss: 0.5844 - val_Recall: 0.8829 - val_loss: 0.2106
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8822 - loss: 0.5651 - val_Recall: 0.8829 - val_loss: 0.2148
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9045 - loss: 0.5323 - val_Recall: 0.8829 - val_loss: 0.2109
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8755 - loss: 0.5814 - val_Recall: 0.8829 - val_loss: 0.2056
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8907 - loss: 0.5507 - val_Recall: 0.8829 - val_loss: 0.2069
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8788 - loss: 0.5401 - val_Recall: 0.8829 - val_loss: 0.1983
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8882 - loss: 0.5583 - val_Recall: 0.8829 - val_loss: 0.1956
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8789 - loss: 0.5341 - val_Recall: 0.8829 - val_loss: 0.1932

print("Time taken in seconds ",end-start)

Time taken in seconds  12.271375179290771

plot(history,'loss')

plot(history,'Recall')

model_9_train_perf = model_performance_classification(model_9, x_train, y_train) #threshold = 0.5
print("Model 8 Train Performance")
print(model_9_train_perf)
print("-"*50)
print("Model 9 Validation Performance")
model_9_val_perf = model_performance_classification(model_9,x_val,y_val)
print(model_9_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 167us/step
Model 8 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.968812  0.938972   0.826937  0.873239
--------------------------------------------------
Model 9 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 180us/step
   Accuracy    Recall  Precision  F1 Score
0   0.96425  0.925957   0.809573  0.856777

#let's get classification report on y train and y val for Model 9
y_train_pred_9 = model_9.predict(x_train)
y_val_pred_9 = model_9.predict(x_val)
print("Classification Report - Train data Model_9",end="\n\n")
cr_train_model_9 = classification_report(y_train,y_train_pred_9>0.5)
print(cr_train_model_9)
print("Classification Report - Validation data Model_9",end="\n\n")
cr_val_model_9 = classification_report(y_val,y_val_pred_9>0.5)
print(cr_val_model_9)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 165us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 189us/step
Classification Report - Train data Model_9

              precision    recall  f1-score   support

           0       0.99      0.97      0.98     15112
           1       0.66      0.91      0.76       888

    accuracy                           0.97     16000
   macro avg       0.83      0.94      0.87     16000
weighted avg       0.98      0.97      0.97     16000

Classification Report - Validation data Model_9

              precision    recall  f1-score   support

           0       0.99      0.97      0.98      3778
           1       0.63      0.88      0.73       222

    accuracy                           0.96      4000
   macro avg       0.81      0.93      0.86      4000
weighted avg       0.97      0.96      0.97      4000

# Predict probabilities using confusion matrix 
y_pred = model_9.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 189us/step

results.loc[9] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[1e-5, "-"],"Xavier - CW ","dropout(0.3) + Batch Norm",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
model_10 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_10.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1],kernel_initializer='he_normal'))
model_10.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_10.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_10.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_10.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_10.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_10.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_10.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_10.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization             │ (None, 128)            │           512 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_1           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_2           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 18,753 (73.25 KB)

 Trainable params: 18,241 (71.25 KB)

 Non-trainable params: 512 (2.00 KB)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_10.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

start = time.time()
history = model_10.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, epochs=epochs,class_weight=cw_dict)
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 625us/step - Recall: 0.5208 - loss: 1.6560 - val_Recall: 0.7072 - val_loss: 0.6930
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.6555 - loss: 1.4446 - val_Recall: 0.7793 - val_loss: 0.6734
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.7235 - loss: 1.2980 - val_Recall: 0.8288 - val_loss: 0.6470
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.7922 - loss: 1.2102 - val_Recall: 0.8514 - val_loss: 0.6222
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8333 - loss: 1.1184 - val_Recall: 0.8559 - val_loss: 0.6029
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8218 - loss: 1.1295 - val_Recall: 0.8739 - val_loss: 0.5852
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8698 - loss: 1.0220 - val_Recall: 0.8739 - val_loss: 0.5707
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8829 - loss: 0.9907 - val_Recall: 0.8739 - val_loss: 0.5471
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8560 - loss: 0.9997 - val_Recall: 0.8694 - val_loss: 0.5312
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8787 - loss: 0.9429 - val_Recall: 0.8739 - val_loss: 0.5141
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8703 - loss: 0.9291 - val_Recall: 0.8784 - val_loss: 0.5020
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8714 - loss: 0.9330 - val_Recall: 0.8829 - val_loss: 0.4825
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8838 - loss: 0.8602 - val_Recall: 0.8829 - val_loss: 0.4723
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 451us/step - Recall: 0.8839 - loss: 0.8981 - val_Recall: 0.8874 - val_loss: 0.4437
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9023 - loss: 0.8260 - val_Recall: 0.8964 - val_loss: 0.4341
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8870 - loss: 0.8222 - val_Recall: 0.8964 - val_loss: 0.4263
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.9140 - loss: 0.7955 - val_Recall: 0.8964 - val_loss: 0.4158
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8952 - loss: 0.8089 - val_Recall: 0.8964 - val_loss: 0.4046
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8940 - loss: 0.8068 - val_Recall: 0.8964 - val_loss: 0.3904
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8915 - loss: 0.7812 - val_Recall: 0.8919 - val_loss: 0.3787
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8959 - loss: 0.7482 - val_Recall: 0.8919 - val_loss: 0.3729
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8632 - loss: 0.7844 - val_Recall: 0.8829 - val_loss: 0.3577
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.9016 - loss: 0.7322 - val_Recall: 0.8829 - val_loss: 0.3509
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.9060 - loss: 0.6965 - val_Recall: 0.8829 - val_loss: 0.3424
Epoch 25/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.8834 - loss: 0.7620 - val_Recall: 0.8829 - val_loss: 0.3328
Epoch 26/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8833 - loss: 0.7320 - val_Recall: 0.8829 - val_loss: 0.3228
Epoch 27/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8840 - loss: 0.7153 - val_Recall: 0.8829 - val_loss: 0.3145
Epoch 28/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8822 - loss: 0.6880 - val_Recall: 0.8829 - val_loss: 0.3114
Epoch 29/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8797 - loss: 0.6876 - val_Recall: 0.8829 - val_loss: 0.3049
Epoch 30/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8823 - loss: 0.7150 - val_Recall: 0.8829 - val_loss: 0.2954
Epoch 31/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9004 - loss: 0.6508 - val_Recall: 0.8829 - val_loss: 0.2879
Epoch 32/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8670 - loss: 0.6666 - val_Recall: 0.8829 - val_loss: 0.2770
Epoch 33/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8993 - loss: 0.6370 - val_Recall: 0.8829 - val_loss: 0.2771
Epoch 34/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.9002 - loss: 0.6169 - val_Recall: 0.8829 - val_loss: 0.2689
Epoch 35/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 504us/step - Recall: 0.8858 - loss: 0.6286 - val_Recall: 0.8829 - val_loss: 0.2621
Epoch 36/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8924 - loss: 0.6171 - val_Recall: 0.8829 - val_loss: 0.2583
Epoch 37/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8681 - loss: 0.6645 - val_Recall: 0.8829 - val_loss: 0.2538
Epoch 38/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8849 - loss: 0.6055 - val_Recall: 0.8829 - val_loss: 0.2500
Epoch 39/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8802 - loss: 0.6158 - val_Recall: 0.8829 - val_loss: 0.2466
Epoch 40/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.9197 - loss: 0.5482 - val_Recall: 0.8784 - val_loss: 0.2395
Epoch 41/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8956 - loss: 0.6082 - val_Recall: 0.8784 - val_loss: 0.2434
Epoch 42/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8924 - loss: 0.5944 - val_Recall: 0.8784 - val_loss: 0.2354
Epoch 43/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8889 - loss: 0.5980 - val_Recall: 0.8784 - val_loss: 0.2261
Epoch 44/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8958 - loss: 0.5788 - val_Recall: 0.8784 - val_loss: 0.2208
Epoch 45/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8788 - loss: 0.5873 - val_Recall: 0.8829 - val_loss: 0.2256
Epoch 46/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.8739 - loss: 0.5943 - val_Recall: 0.8829 - val_loss: 0.2173
Epoch 47/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.8926 - loss: 0.5930 - val_Recall: 0.8829 - val_loss: 0.2178
Epoch 48/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8913 - loss: 0.5674 - val_Recall: 0.8829 - val_loss: 0.2184
Epoch 49/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8956 - loss: 0.5625 - val_Recall: 0.8829 - val_loss: 0.2165
Epoch 50/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8897 - loss: 0.5513 - val_Recall: 0.8829 - val_loss: 0.2091

print("Time taken in seconds ",end-start)

Time taken in seconds  12.212008953094482

plot(history,'loss')

plot(history,'Recall')

model_10_train_perf = model_performance_classification(model_10, x_train, y_train) #threshold = 0.5
print("Model 10 Train Performance")
print(model_10_train_perf)
print("-"*50)
print("Model 10 Validation Performance")
model_10_val_perf = model_performance_classification(model_10,x_val,y_val)
print(model_10_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 168us/step
Model 10 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.956187  0.930699   0.780056  0.836005
--------------------------------------------------
Model 10 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 182us/step
   Accuracy    Recall  Precision  F1 Score
0   0.95275  0.919869    0.76941  0.824612

#let's get classification report on y train and y val for Model 10
y_train_pred_10 = model_10.predict(x_train)
y_val_pred_10 = model_10.predict(x_val)
print("Classification Report - Train data Model_10",end="\n\n")
cr_train_model_10 = classification_report(y_train,y_train_pred_10>0.5)
print(cr_train_model_10)
print("Classification Report - Validation data Model_10",end="\n\n")
cr_val_model_10 = classification_report(y_val,y_val_pred_10>0.5)
print(cr_val_model_10)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 164us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 185us/step
Classification Report - Train data Model_10

              precision    recall  f1-score   support

           0       0.99      0.96      0.98     15112
           1       0.57      0.90      0.70       888

    accuracy                           0.96     16000
   macro avg       0.78      0.93      0.84     16000
weighted avg       0.97      0.96      0.96     16000

Classification Report - Validation data Model_10

              precision    recall  f1-score   support

           0       0.99      0.96      0.97      3778
           1       0.55      0.88      0.67       222

    accuracy                           0.95      4000
   macro avg       0.77      0.92      0.82      4000
weighted avg       0.97      0.95      0.96      4000

# Predict probabilities using confusion matrix 
y_pred = model_10.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 184us/step

results.loc[10] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[1e-5, "-"],"He Normal - CW ","dropout(0.3) + Batch Norm",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
#chanig the weight initializer back to Xavier or glorot_normal
model_11 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_11.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1],kernel_initializer='glorot_normal'))
model_11.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_11.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_11.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_11.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_11.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_11.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_11.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_11.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization             │ (None, 128)            │           512 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_1           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_2           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 18,753 (73.25 KB)

 Trainable params: 18,241 (71.25 KB)

 Non-trainable params: 512 (2.00 KB)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_11.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

#callbacks to reduce osciallation and overfitting
from keras.callbacks import ReduceLROnPlateau, EarlyStopping 
lr_schedule = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=5, verbose=1, min_lr=1e-6) #learnign schedule
early_stop = EarlyStopping(monitor='val_Recall',mode='max', patience=10, restore_best_weights=True)  #early stopping

start = time.time()
history = model_11.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=batch_size, 
                       epochs=epochs,class_weight=cw_dict,callbacks=[lr_schedule,early_stop])
end=time.time()

Epoch 1/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 1s 619us/step - Recall: 0.2812 - loss: 2.2647 - val_Recall: 0.3018 - val_loss: 0.7829 - learning_rate: 1.0000e-05
Epoch 2/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.5273 - loss: 1.7191 - val_Recall: 0.5495 - val_loss: 0.7215 - learning_rate: 1.0000e-05
Epoch 3/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.6522 - loss: 1.5047 - val_Recall: 0.6982 - val_loss: 0.6891 - learning_rate: 1.0000e-05
Epoch 4/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.7554 - loss: 1.2881 - val_Recall: 0.8153 - val_loss: 0.6747 - learning_rate: 1.0000e-05
Epoch 5/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.7987 - loss: 1.2433 - val_Recall: 0.8514 - val_loss: 0.6460 - learning_rate: 1.0000e-05
Epoch 6/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.8683 - loss: 1.0970 - val_Recall: 0.8874 - val_loss: 0.6231 - learning_rate: 1.0000e-05
Epoch 7/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8482 - loss: 1.1173 - val_Recall: 0.8829 - val_loss: 0.6000 - learning_rate: 1.0000e-05
Epoch 8/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 457us/step - Recall: 0.8890 - loss: 1.0433 - val_Recall: 0.8874 - val_loss: 0.5829 - learning_rate: 1.0000e-05
Epoch 9/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.8972 - loss: 0.9818 - val_Recall: 0.8874 - val_loss: 0.5617 - learning_rate: 1.0000e-05
Epoch 10/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9007 - loss: 0.9742 - val_Recall: 0.8874 - val_loss: 0.5493 - learning_rate: 1.0000e-05
Epoch 11/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 451us/step - Recall: 0.9083 - loss: 0.9161 - val_Recall: 0.8919 - val_loss: 0.5216 - learning_rate: 1.0000e-05
Epoch 12/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 451us/step - Recall: 0.9262 - loss: 0.8664 - val_Recall: 0.8919 - val_loss: 0.5076 - learning_rate: 1.0000e-05
Epoch 13/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 452us/step - Recall: 0.9011 - loss: 0.8914 - val_Recall: 0.8919 - val_loss: 0.4808 - learning_rate: 1.0000e-05
Epoch 14/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 455us/step - Recall: 0.9070 - loss: 0.8500 - val_Recall: 0.8964 - val_loss: 0.4771 - learning_rate: 1.0000e-05
Epoch 15/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.9010 - loss: 0.8513 - val_Recall: 0.8874 - val_loss: 0.4464 - learning_rate: 1.0000e-05
Epoch 16/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 456us/step - Recall: 0.8949 - loss: 0.8204 - val_Recall: 0.8874 - val_loss: 0.4390 - learning_rate: 1.0000e-05
Epoch 17/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 460us/step - Recall: 0.9004 - loss: 0.7995 - val_Recall: 0.8919 - val_loss: 0.4279 - learning_rate: 1.0000e-05
Epoch 18/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 463us/step - Recall: 0.9107 - loss: 0.7849 - val_Recall: 0.8874 - val_loss: 0.4061 - learning_rate: 1.0000e-05
Epoch 19/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 458us/step - Recall: 0.8903 - loss: 0.7941 - val_Recall: 0.8874 - val_loss: 0.3973 - learning_rate: 1.0000e-05
Epoch 20/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 454us/step - Recall: 0.8880 - loss: 0.7630 - val_Recall: 0.8829 - val_loss: 0.3757 - learning_rate: 1.0000e-05
Epoch 21/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 453us/step - Recall: 0.9072 - loss: 0.7247 - val_Recall: 0.8829 - val_loss: 0.3673 - learning_rate: 1.0000e-05
Epoch 22/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 459us/step - Recall: 0.8999 - loss: 0.7403 - val_Recall: 0.8829 - val_loss: 0.3692 - learning_rate: 1.0000e-05
Epoch 23/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 457us/step - Recall: 0.8900 - loss: 0.7377 - val_Recall: 0.8829 - val_loss: 0.3512 - learning_rate: 1.0000e-05
Epoch 24/50
500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 459us/step - Recall: 0.8983 - loss: 0.7156 - val_Recall: 0.8829 - val_loss: 0.3378 - learning_rate: 1.0000e-05

print("Time taken in seconds ",end-start)

Time taken in seconds  6.189470052719116

plot(history,'loss')

plot(history,'Recall')

model_11_train_perf = model_performance_classification(model_11, x_train, y_train) #threshold = 0.5
print("Model 11 Train Performance")
print(model_11_train_perf)
print("-"*50)
print("Model 11 Validation Performance")
model_11_val_perf = model_performance_classification(model_11,x_val,y_val)
print(model_11_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 169us/step
Model 11 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.844375  0.875747    0.62253  0.652296
--------------------------------------------------
Model 11 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 180us/step
   Accuracy    Recall  Precision  F1 Score
0     0.837  0.864953   0.616544  0.642617

#let's get classification report on y train and y val for Model 11
y_train_pred_11 = model_11.predict(x_train)
y_val_pred_11 = model_11.predict(x_val)
print("Classification Report - Train data Model_11",end="\n\n")
cr_train_model_11 = classification_report(y_train,y_train_pred_11>0.5)
print(cr_train_model_11)
print("Classification Report - Validation data Model_11",end="\n\n")
cr_val_model_11 = classification_report(y_val,y_val_pred_11>0.5)
print(cr_val_model_11)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 167us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 187us/step
Classification Report - Train data Model_11

              precision    recall  f1-score   support

           0       0.99      0.84      0.91     15112
           1       0.25      0.91      0.39       888

    accuracy                           0.84     16000
   macro avg       0.62      0.88      0.65     16000
weighted avg       0.95      0.84      0.88     16000

Classification Report - Validation data Model_11

              precision    recall  f1-score   support

           0       0.99      0.83      0.91      3778
           1       0.24      0.90      0.38       222

    accuracy                           0.84      4000
   macro avg       0.62      0.86      0.64      4000
weighted avg       0.95      0.84      0.88      4000

# Predict probabilities using confusion matrix 
y_pred = model_11.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 190us/step

results.loc[11] = [3,[128,64,64],["relu","relu","relu"],50,32,"adam",[1e-5, "-"],"xavier - CW ","dropout(0.3) + Batch Norm + Callbacks[lr_schedule, early_stop]",history.history["loss"][-1],history.history["val_loss"][-1],history.history["Recall"][-1],history.history["val_Recall"][-1],round(end-start,2)]

results

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
#chanig the weight initializer back to Xavier or glorot_normal
model_12 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_12.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1],kernel_initializer='glorot_normal'))
model_12.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_12.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_12.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_12.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_12.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_12.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_12.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_12.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization             │ (None, 128)            │           512 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_1           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_2           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 18,753 (73.25 KB)

 Trainable params: 18,241 (71.25 KB)

 Non-trainable params: 512 (2.00 KB)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_12.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

#callbacks to reduce osciallation and overfitting

lr_schedule = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=5, verbose=1, min_lr=1e-6) #learnign schedule


start = time.time()
history = model_12.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=64, 
                       epochs=epochs,class_weight=cw_dict,callbacks=[lr_schedule])
end=time.time()

Epoch 1/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 1s 879us/step - Recall: 0.5377 - loss: 1.9114 - val_Recall: 0.6937 - val_loss: 0.8911 - learning_rate: 1.0000e-05
Epoch 2/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 552us/step - Recall: 0.6536 - loss: 1.6402 - val_Recall: 0.7928 - val_loss: 0.8866 - learning_rate: 1.0000e-05
Epoch 3/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.7294 - loss: 1.4604 - val_Recall: 0.8784 - val_loss: 0.8528 - learning_rate: 1.0000e-05
Epoch 4/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.7713 - loss: 1.3327 - val_Recall: 0.8964 - val_loss: 0.8126 - learning_rate: 1.0000e-05
Epoch 5/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.8010 - loss: 1.2428 - val_Recall: 0.9234 - val_loss: 0.7849 - learning_rate: 1.0000e-05
Epoch 6/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.8220 - loss: 1.1750 - val_Recall: 0.9279 - val_loss: 0.7682 - learning_rate: 1.0000e-05
Epoch 7/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 549us/step - Recall: 0.8547 - loss: 1.1040 - val_Recall: 0.9234 - val_loss: 0.7381 - learning_rate: 1.0000e-05
Epoch 8/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 546us/step - Recall: 0.8745 - loss: 1.0874 - val_Recall: 0.9189 - val_loss: 0.7168 - learning_rate: 1.0000e-05
Epoch 9/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.8942 - loss: 1.0094 - val_Recall: 0.9234 - val_loss: 0.7003 - learning_rate: 1.0000e-05
Epoch 10/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.8992 - loss: 0.9889 - val_Recall: 0.9189 - val_loss: 0.6812 - learning_rate: 1.0000e-05
Epoch 11/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.8972 - loss: 0.9886 - val_Recall: 0.9189 - val_loss: 0.6626 - learning_rate: 1.0000e-05
Epoch 12/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9100 - loss: 0.9649 - val_Recall: 0.9279 - val_loss: 0.6440 - learning_rate: 1.0000e-05
Epoch 13/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9080 - loss: 0.9110 - val_Recall: 0.9279 - val_loss: 0.6228 - learning_rate: 1.0000e-05
Epoch 14/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 540us/step - Recall: 0.9098 - loss: 0.8882 - val_Recall: 0.9279 - val_loss: 0.6125 - learning_rate: 1.0000e-05
Epoch 15/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.8935 - loss: 0.8951 - val_Recall: 0.9279 - val_loss: 0.5838 - learning_rate: 1.0000e-05
Epoch 16/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9098 - loss: 0.8827 - val_Recall: 0.9279 - val_loss: 0.5755 - learning_rate: 1.0000e-05
Epoch 17/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9028 - loss: 0.8858 - val_Recall: 0.9324 - val_loss: 0.5597 - learning_rate: 1.0000e-05
Epoch 18/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9283 - loss: 0.8102 - val_Recall: 0.9279 - val_loss: 0.5509 - learning_rate: 1.0000e-05
Epoch 19/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.8973 - loss: 0.8228 - val_Recall: 0.9189 - val_loss: 0.5329 - learning_rate: 1.0000e-05
Epoch 20/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.8984 - loss: 0.8636 - val_Recall: 0.9144 - val_loss: 0.5204 - learning_rate: 1.0000e-05
Epoch 21/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9130 - loss: 0.8144 - val_Recall: 0.9144 - val_loss: 0.5069 - learning_rate: 1.0000e-05
Epoch 22/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.9222 - loss: 0.7693 - val_Recall: 0.9144 - val_loss: 0.4971 - learning_rate: 1.0000e-05
Epoch 23/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 552us/step - Recall: 0.9057 - loss: 0.8046 - val_Recall: 0.9099 - val_loss: 0.4786 - learning_rate: 1.0000e-05
Epoch 24/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 540us/step - Recall: 0.9227 - loss: 0.7581 - val_Recall: 0.9099 - val_loss: 0.4726 - learning_rate: 1.0000e-05
Epoch 25/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.8978 - loss: 0.7526 - val_Recall: 0.9099 - val_loss: 0.4612 - learning_rate: 1.0000e-05
Epoch 26/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 635us/step - Recall: 0.9036 - loss: 0.7821 - val_Recall: 0.9099 - val_loss: 0.4465 - learning_rate: 1.0000e-05
Epoch 27/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 546us/step - Recall: 0.9120 - loss: 0.7368 - val_Recall: 0.9099 - val_loss: 0.4407 - learning_rate: 1.0000e-05
Epoch 28/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.9103 - loss: 0.7214 - val_Recall: 0.9099 - val_loss: 0.4337 - learning_rate: 1.0000e-05
Epoch 29/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.9047 - loss: 0.7468 - val_Recall: 0.9099 - val_loss: 0.4261 - learning_rate: 1.0000e-05
Epoch 30/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.9009 - loss: 0.7114 - val_Recall: 0.9099 - val_loss: 0.4106 - learning_rate: 1.0000e-05
Epoch 31/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 545us/step - Recall: 0.9277 - loss: 0.6829 - val_Recall: 0.9099 - val_loss: 0.4061 - learning_rate: 1.0000e-05
Epoch 32/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9071 - loss: 0.6799 - val_Recall: 0.9099 - val_loss: 0.3953 - learning_rate: 1.0000e-05
Epoch 33/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9220 - loss: 0.6711 - val_Recall: 0.9099 - val_loss: 0.3922 - learning_rate: 1.0000e-05
Epoch 34/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.9105 - loss: 0.6765 - val_Recall: 0.9099 - val_loss: 0.3731 - learning_rate: 1.0000e-05
Epoch 35/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.9204 - loss: 0.6538 - val_Recall: 0.9099 - val_loss: 0.3761 - learning_rate: 1.0000e-05
Epoch 36/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.9011 - loss: 0.6830 - val_Recall: 0.9099 - val_loss: 0.3614 - learning_rate: 1.0000e-05
Epoch 37/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.8930 - loss: 0.6673 - val_Recall: 0.9099 - val_loss: 0.3528 - learning_rate: 1.0000e-05
Epoch 38/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.8866 - loss: 0.6781 - val_Recall: 0.9099 - val_loss: 0.3472 - learning_rate: 1.0000e-05
Epoch 39/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.9158 - loss: 0.6336 - val_Recall: 0.9099 - val_loss: 0.3439 - learning_rate: 1.0000e-05
Epoch 40/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 552us/step - Recall: 0.9252 - loss: 0.6068 - val_Recall: 0.9054 - val_loss: 0.3332 - learning_rate: 1.0000e-05
Epoch 41/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.8849 - loss: 0.6485 - val_Recall: 0.9009 - val_loss: 0.3243 - learning_rate: 1.0000e-05
Epoch 42/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9016 - loss: 0.6071 - val_Recall: 0.9009 - val_loss: 0.3220 - learning_rate: 1.0000e-05
Epoch 43/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.8907 - loss: 0.6358 - val_Recall: 0.9054 - val_loss: 0.3160 - learning_rate: 1.0000e-05
Epoch 44/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.9070 - loss: 0.6084 - val_Recall: 0.9054 - val_loss: 0.3112 - learning_rate: 1.0000e-05
Epoch 45/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9004 - loss: 0.6103 - val_Recall: 0.9009 - val_loss: 0.3055 - learning_rate: 1.0000e-05
Epoch 46/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 546us/step - Recall: 0.9021 - loss: 0.6079 - val_Recall: 0.9009 - val_loss: 0.3008 - learning_rate: 1.0000e-05
Epoch 47/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 543us/step - Recall: 0.9112 - loss: 0.5849 - val_Recall: 0.8964 - val_loss: 0.2900 - learning_rate: 1.0000e-05
Epoch 48/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 548us/step - Recall: 0.9019 - loss: 0.5767 - val_Recall: 0.8964 - val_loss: 0.2888 - learning_rate: 1.0000e-05
Epoch 49/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 549us/step - Recall: 0.8854 - loss: 0.6155 - val_Recall: 0.8964 - val_loss: 0.2821 - learning_rate: 1.0000e-05
Epoch 50/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 549us/step - Recall: 0.8937 - loss: 0.5899 - val_Recall: 0.8964 - val_loss: 0.2820 - learning_rate: 1.0000e-05

print("Time taken in seconds ",end-start)

Time taken in seconds  7.67601203918457

plot(history,'loss')

plot(history,'Recall')

model_12_train_perf = model_performance_classification(model_12, x_train, y_train) #threshold = 0.5
print("Model 12 Train Performance")
print(model_12_train_perf)
print("-"*50)
print("Model 12 Validation Performance")
model_12_val_perf = model_performance_classification(model_12,x_val,y_val)
print(model_12_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 170us/step
Model 12 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.946875  0.929479   0.753247  0.813266
--------------------------------------------------
Model 12 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 183us/step
   Accuracy    Recall  Precision  F1 Score
0     0.943  0.921067   0.743089  0.802432

#let's get classification report on y train and y val for Model 12
y_train_pred_12 = model_12.predict(x_train)
y_val_pred_12 = model_12.predict(x_val)
print("Classification Report - Train data Model_12",end="\n\n")
cr_train_model_12 = classification_report(y_train,y_train_pred_12>0.5)
print(cr_train_model_12)
print("Classification Report - Validation data Model_12",end="\n\n")
cr_val_model_12 = classification_report(y_val,y_val_pred_12>0.5)
print(cr_val_model_12)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 168us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 191us/step
Classification Report - Train data Model_12

              precision    recall  f1-score   support

           0       0.99      0.95      0.97     15112
           1       0.51      0.91      0.66       888

    accuracy                           0.95     16000
   macro avg       0.75      0.93      0.81     16000
weighted avg       0.97      0.95      0.95     16000

Classification Report - Validation data Model_12

              precision    recall  f1-score   support

           0       0.99      0.95      0.97      3778
           1       0.49      0.90      0.64       222

    accuracy                           0.94      4000
   macro avg       0.74      0.92      0.80      4000
weighted avg       0.97      0.94      0.95      4000

# Predict probabilities using confusion matrix 
y_pred = model_12.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.5:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 193us/step

#clear current keras session
tf.keras.backend.clear_session()

# initializing the neural network
#chanig the weight initializer back to Xavier or glorot_normal
model_13 = Sequential()

# Adding a dense layer with ReLu activation function, with the input dimension determined by the shape of the training data.
model_13.add(Dense(128, activation = 'relu', input_dim = x_train.shape[1],kernel_initializer='glorot_normal'))
model_13.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_13.add(Dropout(0.3))  # Adding 30% Dropout after first hidden layer
model_13.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_13.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_13.add(Dense(64, activation = 'relu'))  #Adding a dense layer with ReLu activation function
model_13.add(BatchNormalization()) # Adding Batch Normalization after Dense Layer
model_13.add(Dense(1,activation = 'sigmoid')) #adding output layer with sigmoid as the output is binary 1 or 0

/opt/anaconda3/lib/python3.12/site-packages/keras/src/layers/core/dense.py:93: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

model_13.summary()

Model: "sequential"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ dense (Dense)                   │ (None, 128)            │         5,248 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization             │ (None, 128)            │           512 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dropout (Dropout)               │ (None, 128)            │             0 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_1 (Dense)                 │ (None, 64)             │         8,256 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_1           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_2 (Dense)                 │ (None, 64)             │         4,160 │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ batch_normalization_2           │ (None, 64)             │           256 │
│ (BatchNormalization)            │                        │               │
├─────────────────────────────────┼────────────────────────┼───────────────┤
│ dense_3 (Dense)                 │ (None, 1)              │            65 │
└─────────────────────────────────┴────────────────────────┴───────────────┘

 Total params: 18,753 (73.25 KB)

 Trainable params: 18,241 (71.25 KB)

 Non-trainable params: 512 (2.00 KB)

#Adam Optiomizer 
lr = 1e-5
optimizer = keras.optimizers.Adam(learning_rate=lr)    # defining SGD as the optimizer to be used
model_13.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["Recall"])

#callbacks to reduce osciallation and overfitting

lr_schedule = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=5, verbose=1, min_lr=1e-6) #learnign schedule


start = time.time()
history = model_13.fit(x_train, y_train, validation_data=(x_val,y_val) , batch_size=64, 
                       epochs=epochs,class_weight=cw_dict,callbacks=[lr_schedule])
end=time.time()

Epoch 1/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 1s 852us/step - Recall: 0.4774 - loss: 1.7069 - val_Recall: 0.6261 - val_loss: 0.7153 - learning_rate: 1.0000e-05
Epoch 2/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 526us/step - Recall: 0.6269 - loss: 1.4742 - val_Recall: 0.7477 - val_loss: 0.6806 - learning_rate: 1.0000e-05
Epoch 3/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 520us/step - Recall: 0.7155 - loss: 1.3133 - val_Recall: 0.8423 - val_loss: 0.6618 - learning_rate: 1.0000e-05
Epoch 4/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 517us/step - Recall: 0.7981 - loss: 1.2215 - val_Recall: 0.8829 - val_loss: 0.6404 - learning_rate: 1.0000e-05
Epoch 5/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 517us/step - Recall: 0.8406 - loss: 1.1414 - val_Recall: 0.8919 - val_loss: 0.6215 - learning_rate: 1.0000e-05
Epoch 6/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 517us/step - Recall: 0.8630 - loss: 1.0986 - val_Recall: 0.8874 - val_loss: 0.6118 - learning_rate: 1.0000e-05
Epoch 7/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 518us/step - Recall: 0.8772 - loss: 1.0879 - val_Recall: 0.8964 - val_loss: 0.6070 - learning_rate: 1.0000e-05
Epoch 8/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 526us/step - Recall: 0.8908 - loss: 1.0408 - val_Recall: 0.9009 - val_loss: 0.5853 - learning_rate: 1.0000e-05
Epoch 9/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 557us/step - Recall: 0.8891 - loss: 0.9865 - val_Recall: 0.9054 - val_loss: 0.5687 - learning_rate: 1.0000e-05
Epoch 10/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 550us/step - Recall: 0.9045 - loss: 0.9607 - val_Recall: 0.9099 - val_loss: 0.5549 - learning_rate: 1.0000e-05
Epoch 11/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 536us/step - Recall: 0.9035 - loss: 0.9573 - val_Recall: 0.9099 - val_loss: 0.5421 - learning_rate: 1.0000e-05
Epoch 12/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 536us/step - Recall: 0.9292 - loss: 0.9057 - val_Recall: 0.9189 - val_loss: 0.5264 - learning_rate: 1.0000e-05
Epoch 13/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 519us/step - Recall: 0.9237 - loss: 0.9133 - val_Recall: 0.9234 - val_loss: 0.5176 - learning_rate: 1.0000e-05
Epoch 14/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 695us/step - Recall: 0.9306 - loss: 0.8960 - val_Recall: 0.9234 - val_loss: 0.5109 - learning_rate: 1.0000e-05
Epoch 15/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 568us/step - Recall: 0.9188 - loss: 0.8707 - val_Recall: 0.9234 - val_loss: 0.4931 - learning_rate: 1.0000e-05
Epoch 16/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 529us/step - Recall: 0.9215 - loss: 0.8733 - val_Recall: 0.9234 - val_loss: 0.4864 - learning_rate: 1.0000e-05
Epoch 17/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - Recall: 0.9230 - loss: 0.8532 - val_Recall: 0.9234 - val_loss: 0.4704 - learning_rate: 1.0000e-05
Epoch 18/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 550us/step - Recall: 0.9171 - loss: 0.8296 - val_Recall: 0.9234 - val_loss: 0.4618 - learning_rate: 1.0000e-05
Epoch 19/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 551us/step - Recall: 0.9249 - loss: 0.8135 - val_Recall: 0.9234 - val_loss: 0.4496 - learning_rate: 1.0000e-05
Epoch 20/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 534us/step - Recall: 0.9237 - loss: 0.8023 - val_Recall: 0.9234 - val_loss: 0.4403 - learning_rate: 1.0000e-05
Epoch 21/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9267 - loss: 0.7885 - val_Recall: 0.9234 - val_loss: 0.4288 - learning_rate: 1.0000e-05
Epoch 22/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 555us/step - Recall: 0.9045 - loss: 0.8147 - val_Recall: 0.9234 - val_loss: 0.4209 - learning_rate: 1.0000e-05
Epoch 23/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 546us/step - Recall: 0.9162 - loss: 0.7921 - val_Recall: 0.9234 - val_loss: 0.4139 - learning_rate: 1.0000e-05
Epoch 24/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.9151 - loss: 0.7717 - val_Recall: 0.9234 - val_loss: 0.4002 - learning_rate: 1.0000e-05
Epoch 25/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 527us/step - Recall: 0.9080 - loss: 0.7666 - val_Recall: 0.9234 - val_loss: 0.3976 - learning_rate: 1.0000e-05
Epoch 26/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 533us/step - Recall: 0.9177 - loss: 0.7684 - val_Recall: 0.9234 - val_loss: 0.3879 - learning_rate: 1.0000e-05
Epoch 27/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - Recall: 0.9217 - loss: 0.7429 - val_Recall: 0.9234 - val_loss: 0.3792 - learning_rate: 1.0000e-05
Epoch 28/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 551us/step - Recall: 0.9063 - loss: 0.7674 - val_Recall: 0.9234 - val_loss: 0.3723 - learning_rate: 1.0000e-05
Epoch 29/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 555us/step - Recall: 0.9156 - loss: 0.7189 - val_Recall: 0.9234 - val_loss: 0.3614 - learning_rate: 1.0000e-05
Epoch 30/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - Recall: 0.9078 - loss: 0.7119 - val_Recall: 0.9234 - val_loss: 0.3524 - learning_rate: 1.0000e-05
Epoch 31/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - Recall: 0.9098 - loss: 0.7160 - val_Recall: 0.9189 - val_loss: 0.3467 - learning_rate: 1.0000e-05
Epoch 32/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 525us/step - Recall: 0.9059 - loss: 0.7227 - val_Recall: 0.9189 - val_loss: 0.3402 - learning_rate: 1.0000e-05
Epoch 33/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 529us/step - Recall: 0.9140 - loss: 0.6872 - val_Recall: 0.9189 - val_loss: 0.3370 - learning_rate: 1.0000e-05
Epoch 34/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 541us/step - Recall: 0.9094 - loss: 0.6732 - val_Recall: 0.9144 - val_loss: 0.3215 - learning_rate: 1.0000e-05
Epoch 35/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 534us/step - Recall: 0.9057 - loss: 0.6787 - val_Recall: 0.9144 - val_loss: 0.3210 - learning_rate: 1.0000e-05
Epoch 36/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 525us/step - Recall: 0.9198 - loss: 0.6622 - val_Recall: 0.9144 - val_loss: 0.3117 - learning_rate: 1.0000e-05
Epoch 37/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 552us/step - Recall: 0.9118 - loss: 0.6672 - val_Recall: 0.9144 - val_loss: 0.3067 - learning_rate: 1.0000e-05
Epoch 38/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 547us/step - Recall: 0.9036 - loss: 0.6719 - val_Recall: 0.9144 - val_loss: 0.3031 - learning_rate: 1.0000e-05
Epoch 39/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9054 - loss: 0.6753 - val_Recall: 0.9144 - val_loss: 0.2977 - learning_rate: 1.0000e-05
Epoch 40/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 668us/step - Recall: 0.9113 - loss: 0.6444 - val_Recall: 0.9009 - val_loss: 0.2906 - learning_rate: 1.0000e-05
Epoch 41/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 539us/step - Recall: 0.9227 - loss: 0.6169 - val_Recall: 0.9054 - val_loss: 0.2877 - learning_rate: 1.0000e-05
Epoch 42/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 542us/step - Recall: 0.9226 - loss: 0.6123 - val_Recall: 0.9054 - val_loss: 0.2787 - learning_rate: 1.0000e-05
Epoch 43/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 537us/step - Recall: 0.9021 - loss: 0.6388 - val_Recall: 0.9054 - val_loss: 0.2776 - learning_rate: 1.0000e-05
Epoch 44/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 549us/step - Recall: 0.9169 - loss: 0.6237 - val_Recall: 0.9009 - val_loss: 0.2702 - learning_rate: 1.0000e-05
Epoch 45/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 535us/step - Recall: 0.9182 - loss: 0.5998 - val_Recall: 0.9009 - val_loss: 0.2633 - learning_rate: 1.0000e-05
Epoch 46/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 520us/step - Recall: 0.9138 - loss: 0.5913 - val_Recall: 0.9009 - val_loss: 0.2619 - learning_rate: 1.0000e-05
Epoch 47/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 518us/step - Recall: 0.9015 - loss: 0.6107 - val_Recall: 0.9054 - val_loss: 0.2575 - learning_rate: 1.0000e-05
Epoch 48/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 521us/step - Recall: 0.9036 - loss: 0.6250 - val_Recall: 0.9054 - val_loss: 0.2543 - learning_rate: 1.0000e-05
Epoch 49/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 560us/step - Recall: 0.8869 - loss: 0.6151 - val_Recall: 0.8964 - val_loss: 0.2458 - learning_rate: 1.0000e-05
Epoch 50/50
250/250 ━━━━━━━━━━━━━━━━━━━━ 0s 544us/step - Recall: 0.9052 - loss: 0.6044 - val_Recall: 0.9009 - val_loss: 0.2409 - learning_rate: 1.0000e-05

print("Time taken in seconds ",end-start)

Time taken in seconds  7.6471638679504395

plot(history,'loss')

plot(history,'Recall')

model_13_train_perf = model_performance_classification(model_13, x_train, y_train) #threshold = 0.5
print("Model 13 Train Performance")
print(model_13_train_perf)
print("-"*50)
print("Model 13 Validation Performance")
model_13_val_perf = model_performance_classification(model_13,x_val,y_val)
print(model_13_val_perf)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 183us/step
Model 13 Train Performance
   Accuracy    Recall  Precision  F1 Score
0  0.953063  0.932224     0.7704  0.828555
--------------------------------------------------
Model 13 Validation Performance
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 185us/step
   Accuracy    Recall  Precision  F1 Score
0    0.9515  0.927687   0.765785  0.823603

#let's get classification report on y train and y val for Model 13
y_train_pred_13 = model_13.predict(x_train)
y_val_pred_13 = model_13.predict(x_val)
print("Classification Report - Train data Model_13",end="\n\n")
cr_train_model_13 = classification_report(y_train,y_train_pred_13>0.3)
print(cr_train_model_13)
print("Classification Report - Validation data Model_13",end="\n\n")
cr_val_model_13 = classification_report(y_val,y_val_pred_13>0.3)
print(cr_val_model_13)

500/500 ━━━━━━━━━━━━━━━━━━━━ 0s 188us/step
125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 198us/step
Classification Report - Train data Model_13

              precision    recall  f1-score   support

           0       1.00      0.86      0.92     15112
           1       0.28      0.93      0.43       888

    accuracy                           0.86     16000
   macro avg       0.64      0.90      0.68     16000
weighted avg       0.96      0.86      0.90     16000

Classification Report - Validation data Model_13

              precision    recall  f1-score   support

           0       1.00      0.85      0.92      3778
           1       0.26      0.93      0.41       222

    accuracy                           0.85      4000
   macro avg       0.63      0.89      0.66      4000
weighted avg       0.95      0.85      0.89      4000

# Predict probabilities using confusion matrix 
y_pred = model_13.predict(x_val)
for i in range(len(y_val)):
    if y_pred[i]>0.3:
        y_pred[i]=1 
    else:
        y_pred[i]=0

cm = confusion_matrix(y_val,y_pred)
labels = ['True Negative','False Positive','False Negative','True Positive']
make_confusion_matrix(cm,group_names=labels,cmap = "Blues")

125/125 ━━━━━━━━━━━━━━━━━━━━ 0s 279us/step

# training performance comparison

models_train_comp_df = pd.concat(
    [
        model_0_train_perf.T,
        model_1_train_perf.T,
        model_2_train_perf.T,
        model_3_train_perf.T,
        model_4_train_perf.T,
        model_5_train_perf.T,
        model_6_train_perf.T,
        model_7_train_perf.T,
        model_8_train_perf.T,
        model_9_train_perf.T,
        model_10_train_perf.T,
        model_11_train_perf.T,
        model_12_train_perf.T,
        model_13_train_perf.T
        
    ],
    axis=1,
)
models_train_comp_df.columns = [
    "Model 0",
    "Model 1",
    "Model 2",
    "Model 3",
    "Model 4",
    "Model 5",
    "Model 6",
    "Model 7",
    "Model 8",
    "Model 9",
    "Model 10",
    "Model 11",
    "Model 12",
    "Model 13",
    
    
]
print("Training set performance comparison:")
models_train_comp_df

Training set performance comparison:

# Validation performance comparison

models_val_comp_df = pd.concat(
    [
        model_0_val_perf.T,
        model_1_val_perf.T,
        model_2_val_perf.T,
        model_3_val_perf.T,
        model_4_val_perf.T,
        model_5_val_perf.T,
        model_6_val_perf.T,
        model_7_val_perf.T,
        model_8_val_perf.T,
        model_9_val_perf.T,
        model_10_val_perf.T,
        model_11_val_perf.T,
        model_12_val_perf.T,
        model_13_val_perf.T
        
    ],
    axis=1,
)
models_val_comp_df.columns = [
    "Model 0",
    "Model 1",
    "Model 2",
    "Model 3",
    "Model 4",
    "Model 5",
    "Model 6",
    "Model 7",
    "Model 8",
    "Model 9",
    "Model 10",
    "Model 11",
    "Model 12",
    "Model 13"
    
    
]
print("Validation set performance comparison:")
models_val_comp_df

Validation set performance comparison:

best_model = model_6

best_model_test_perf = model_performance_classification(best_model,X_test,y_test)
best_model_test_perf

157/157 ━━━━━━━━━━━━━━━━━━━━ 0s 294us/step

y_test_pred_best = best_model.predict(X_test)

cr_test_best_model = classification_report(y_test, y_test_pred_best>0.5) # Checking the classification report of best model on test data.
print(cr_test_best_model)

157/157 ━━━━━━━━━━━━━━━━━━━━ 0s 252us/step
              precision    recall  f1-score   support

           0       0.99      0.99      0.99      4718
           1       0.78      0.87      0.82       282

    accuracy                           0.98      5000
   macro avg       0.89      0.93      0.91      5000
weighted avg       0.98      0.98      0.98      5000

	count	mean	std	min	25%	50%	75%	max
V1	19982.0	-0.271996	3.441625	-11.876451	-2.737146	-0.747917	1.840112	15.493002
V2	19982.0	0.440430	3.150784	-12.319951	-1.640674	0.471536	2.543967	13.089269
V3	20000.0	2.484699	3.388963	-10.708139	0.206860	2.255786	4.566165	17.090919
V4	20000.0	-0.083152	3.431595	-15.082052	-2.347660	-0.135241	2.130615	13.236381
V5	20000.0	-0.053752	2.104801	-8.603361	-1.535607	-0.101952	1.340480	8.133797
V6	20000.0	-0.995443	2.040970	-10.227147	-2.347238	-1.000515	0.380330	6.975847
V7	20000.0	-0.879325	1.761626	-7.949681	-2.030926	-0.917179	0.223695	8.006091
V8	20000.0	-0.548195	3.295756	-15.657561	-2.642665	-0.389085	1.722965	11.679495
V9	20000.0	-0.016808	2.160568	-8.596313	-1.494973	-0.067597	1.409203	8.137580
V10	20000.0	-0.012998	2.193201	-9.853957	-1.411212	0.100973	1.477045	8.108472
V11	20000.0	-1.895393	3.124322	-14.832058	-3.922404	-1.921237	0.118906	11.826433
V12	20000.0	1.604825	2.930454	-12.948007	-0.396514	1.507841	3.571454	15.080698
V13	20000.0	1.580486	2.874658	-13.228247	-0.223545	1.637185	3.459886	15.419616
V14	20000.0	-0.950632	1.789651	-7.738593	-2.170741	-0.957163	0.270677	5.670664
V15	20000.0	-2.414993	3.354974	-16.416606	-4.415322	-2.382617	-0.359052	12.246455
V16	20000.0	-2.925225	4.221717	-20.374158	-5.634240	-2.682705	-0.095046	13.583212
V17	20000.0	-0.134261	3.345462	-14.091184	-2.215611	-0.014580	2.068751	16.756432
V18	20000.0	1.189347	2.592276	-11.643994	-0.403917	0.883398	2.571770	13.179863
V19	20000.0	1.181808	3.396925	-13.491784	-1.050168	1.279061	3.493299	13.237742
V20	20000.0	0.023608	3.669477	-13.922659	-2.432953	0.033415	2.512372	16.052339
V21	20000.0	-3.611252	3.567690	-17.956231	-5.930360	-3.532888	-1.265884	13.840473
V22	20000.0	0.951835	1.651547	-10.122095	-0.118127	0.974687	2.025594	7.409856
V23	20000.0	-0.366116	4.031860	-14.866128	-3.098756	-0.262093	2.451750	14.458734
V24	20000.0	1.134389	3.912069	-16.387147	-1.468062	0.969048	3.545975	17.163291
V25	20000.0	-0.002186	2.016740	-8.228266	-1.365178	0.025050	1.397112	8.223389
V26	20000.0	1.873785	3.435137	-11.834271	-0.337863	1.950531	4.130037	16.836410
V27	20000.0	-0.612413	4.368847	-14.904939	-3.652323	-0.884894	2.189177	17.560404
V28	20000.0	-0.883218	1.917713	-9.269489	-2.171218	-0.891073	0.375884	6.527643
V29	20000.0	-0.985625	2.684365	-12.579469	-2.787443	-1.176181	0.629773	10.722055
V30	20000.0	-0.015534	3.005258	-14.796047	-1.867114	0.184346	2.036229	12.505812
V31	20000.0	0.486842	3.461384	-13.722760	-1.817772	0.490304	2.730688	17.255090
V32	20000.0	0.303799	5.500400	-19.876502	-3.420469	0.052073	3.761722	23.633187
V33	20000.0	0.049825	3.575285	-16.898353	-2.242857	-0.066249	2.255134	16.692486
V34	20000.0	-0.462702	3.183841	-17.985094	-2.136984	-0.255008	1.436935	14.358213
V35	20000.0	2.229620	2.937102	-15.349803	0.336191	2.098633	4.064358	15.291065
V36	20000.0	1.514809	3.800860	-14.833178	-0.943809	1.566526	3.983939	19.329576
V37	20000.0	0.011316	1.788165	-5.478350	-1.255819	-0.128435	1.175533	7.467006
V38	20000.0	-0.344025	3.948147	-17.375002	-2.987638	-0.316849	2.279399	15.289923
V39	20000.0	0.890653	1.753054	-6.438880	-0.272250	0.919261	2.057540	7.759877
V40	20000.0	-0.875630	3.012155	-11.023935	-2.940193	-0.920806	1.119897	10.654265
Target	20000.0	0.055500	0.228959	0.000000	0.000000	0.000000	0.000000	1.000000

	count	mean	std	min	25%	50%	75%	max
V1	4995.0	-0.277622	3.466280	-12.381696	-2.743691	-0.764767	1.831313	13.504352
V2	4994.0	0.397928	3.139562	-10.716179	-1.649211	0.427369	2.444486	14.079073
V3	5000.0	2.551787	3.326607	-9.237940	0.314931	2.260428	4.587000	15.314503
V4	5000.0	-0.048943	3.413937	-14.682446	-2.292694	-0.145753	2.166468	12.140157
V5	5000.0	-0.080120	2.110870	-7.711569	-1.615238	-0.131890	1.341197	7.672835
V6	5000.0	-1.042138	2.005444	-8.924196	-2.368853	-1.048571	0.307555	5.067685
V7	5000.0	-0.907922	1.769017	-8.124230	-2.054259	-0.939695	0.212228	7.616182
V8	5000.0	-0.574592	3.331911	-12.252731	-2.642088	-0.357943	1.712896	10.414722
V9	5000.0	0.030121	2.174139	-6.785495	-1.455712	-0.079891	1.449548	8.850720
V10	5000.0	0.018524	2.145437	-8.170956	-1.353320	0.166292	1.511248	6.598728
V11	5000.0	-2.008615	3.112220	-13.151753	-4.050432	-2.043122	0.044069	9.956400
V12	5000.0	1.576413	2.907401	-8.164048	-0.449674	1.488253	3.562626	12.983644
V13	5000.0	1.622456	2.882892	-11.548209	-0.126012	1.718649	3.464604	12.620041
V14	5000.0	-0.921097	1.803470	-7.813929	-2.110952	-0.896011	0.272324	5.734112
V15	5000.0	-2.452174	3.387041	-15.285768	-4.479072	-2.417131	-0.432943	11.673420
V16	5000.0	-3.018503	4.264407	-20.985779	-5.648343	-2.773763	-0.178105	13.975843
V17	5000.0	-0.103721	3.336513	-13.418281	-2.227683	0.047462	2.111907	19.776592
V18	5000.0	1.195606	2.586403	-12.214016	-0.408850	0.881395	2.604014	13.642235
V19	5000.0	1.210490	3.384662	-14.169635	-1.026394	1.295864	3.526278	12.427997
V20	5000.0	0.138429	3.657171	-13.719620	-2.325454	0.193386	2.539550	13.870565
V21	5000.0	-3.664398	3.577841	-16.340707	-5.944369	-3.662870	-1.329645	11.046925
V22	5000.0	0.961960	1.640414	-6.740239	-0.047728	0.986020	2.029321	7.505291
V23	5000.0	-0.422182	4.056714	-14.422274	-3.162690	-0.279222	2.425911	13.180887
V24	5000.0	1.088841	3.968207	-12.315545	-1.623203	0.912815	3.537195	17.806035
V25	5000.0	0.061235	2.010227	-6.770139	-1.298377	0.076703	1.428491	6.556937
V26	5000.0	1.847261	3.400330	-11.414019	-0.242470	1.917032	4.156106	17.528193
V27	5000.0	-0.552397	4.402947	-13.177038	-3.662591	-0.871982	2.247257	17.290161
V28	5000.0	-0.867678	1.926181	-7.933388	-2.159811	-0.930695	0.420587	7.415659
V29	5000.0	-1.095805	2.655454	-9.987800	-2.861373	-1.340547	0.521843	14.039466
V30	5000.0	-0.118699	3.023292	-12.438434	-1.996743	0.112463	1.946450	10.314976
V31	5000.0	0.468810	3.446324	-11.263271	-1.822421	0.485742	2.779008	12.558928
V32	5000.0	0.232567	5.585628	-17.244168	-3.556267	-0.076694	3.751857	26.539391
V33	5000.0	-0.080115	3.538624	-14.903781	-2.348121	-0.159713	2.099160	13.323517
V34	5000.0	-0.392663	3.166101	-14.699725	-2.009604	-0.171745	1.465402	12.146302
V35	5000.0	2.211205	2.948426	-12.260591	0.321818	2.111750	4.031639	13.489237
V36	5000.0	1.594845	3.774970	-12.735567	-0.866066	1.702964	4.104409	17.116122
V37	5000.0	0.022931	1.785320	-5.079070	-1.240526	-0.110415	1.237522	6.809938
V38	5000.0	-0.405659	3.968936	-15.334533	-2.984480	-0.381162	2.287998	13.064950
V39	5000.0	0.938800	1.716502	-5.451050	-0.208024	0.959152	2.130769	7.182237
V40	5000.0	-0.932406	2.978193	-10.076234	-2.986587	-1.002764	1.079738	8.698460
Target	5000.0	0.056400	0.230716	0.000000	0.000000	0.000000	0.000000	1.000000

	# hidden layers	# neurons - hidden layer	activation function - hidden layer	# epochs	batch size	optimizer	learning rate, momentum	weight initializer	regularization	train loss	validation loss	train recall	validation recall	time (secs)
0	1	[128]	[relu]	50	32	[sgd]	[0.01, -]	-	-	0.040534	0.059940	0.861486	0.851351	6.92
1	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	-	0.028434	0.055182	0.902027	0.869369	7.70
2	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.5)	0.050771	0.056083	0.834459	0.855856	8.08
3	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.3)	0.038127	0.053664	0.880631	0.869369	8.56
4	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	Xavier - CW	dropout(0.3)	0.245499	0.108515	0.925676	0.891892	8.46

	# hidden layers	# neurons - hidden layer	activation function - hidden layer	# epochs	batch size	optimizer	learning rate, momentum	weight initializer	regularization	train loss	validation loss	train recall	validation recall	time (secs)
0	1	[128]	[relu]	50	32	[sgd]	[0.01, -]	-	-	0.040534	0.059940	0.861486	0.851351	6.92
1	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	-	0.028434	0.055182	0.902027	0.869369	7.70
2	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.5)	0.050771	0.056083	0.834459	0.855856	8.08
3	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.3)	0.038127	0.053664	0.880631	0.869369	8.56
4	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	Xavier - CW	dropout(0.3)	0.245499	0.108515	0.925676	0.891892	8.46
5	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, 0.3]	Xavier - CW	dropout(0.3)	0.257260	0.112013	0.921171	0.887387	8.89

	# hidden layers	# neurons - hidden layer	activation function - hidden layer	# epochs	batch size	optimizer	learning rate, momentum	weight initializer	regularization	train loss	validation loss	train recall	validation recall	time (secs)
0	1	[128]	[relu]	50	32	[sgd]	[0.01, -]	-	-	0.040534	0.059940	0.861486	0.851351	6.92
1	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	-	0.028434	0.055182	0.902027	0.869369	7.70
2	2	[128, 64]	[relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.5)	0.050771	0.056083	0.834459	0.855856	8.08
3	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	-	dropout(0.3)	0.038127	0.053664	0.880631	0.869369	8.56
4	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, -]	Xavier - CW	dropout(0.3)	0.245499	0.108515	0.925676	0.891892	8.46
5	3	[128, 64, 64]	[relu, relu, relu]	50	32	sgd	[0.01, 0.3]	Xavier - CW	dropout(0.3)	0.257260	0.112013	0.921171	0.887387	8.89
6	3	[128, 64, 64]	[relu, relu, relu]	50	32	adam	[0.001, -]	Xavier - CW	dropout(0.3)	0.234939	0.143119	0.930180	0.900901	9.93

	Model 0	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12	Model 13
Accuracy	0.992125	0.994437	0.992938	0.993750	0.989812	0.991500	0.987125	0.989313	0.971625	0.968812	0.956187	0.844375	0.946875	0.953063
Recall	0.934884	0.953597	0.941144	0.948463	0.961218	0.962642	0.966685	0.958304	0.938341	0.938972	0.930699	0.875747	0.929479	0.932224
Precision	0.989206	0.992995	0.990906	0.991422	0.943856	0.956836	0.919863	0.941949	0.840087	0.826937	0.780056	0.622530	0.753247	0.770400
F1 Score	0.960243	0.972370	0.964526	0.968834	0.952347	0.959718	0.941851	0.949956	0.881904	0.873239	0.836005	0.652296	0.813266	0.828555

	Model 0	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12	Model 13
Accuracy	0.990000	0.991750	0.990250	0.991500	0.987500	0.988250	0.982500	0.987250	0.970000	0.964250	0.952750	0.837000	0.943000	0.951500
Recall	0.924749	0.934155	0.927002	0.934023	0.942505	0.940782	0.944098	0.940253	0.933241	0.925957	0.919869	0.864953	0.921067	0.927687
Precision	0.977805	0.986035	0.978026	0.983560	0.938786	0.946466	0.900294	0.938395	0.833497	0.809573	0.769410	0.616544	0.743089	0.765785
F1 Score	0.949515	0.958444	0.950888	0.957281	0.940637	0.943603	0.920884	0.939321	0.875704	0.856777	0.824612	0.642617	0.802432	0.823603

	Category	Shape (Row * Col)	Contains Null	Column names having Null values
0	Train	(20000, 41)	36	[V1, V2]
1	Test	(5000, 41)	11	[V1, V2]

Problem Statement¶

Business Context¶

Objective¶

Data Description¶

Installing and Importing the necessary libraries¶

Loading the Data¶

Data Overview¶

Observation on Data Overview¶

Exploratory Data Analysis¶

Univariate analysis¶

Observation on Univariate Analysis¶

Bivariate Analysis¶

Observation on Bivariate Analysis¶

Data Preprocessing¶

Treat Missing Value¶

Model Building¶

Model Evaluation Criterion¶

Utility Functions¶

Initial Model Building (Model 0)¶

Observations¶

Model Performance Improvement¶

Model 1¶

Observations¶

Model 2¶

Observations¶

Model 3¶

Observations¶

Model 4¶

Observations¶

Model 5¶

Observations¶

Model 6¶

Observations¶

Model 7¶

Observations¶

Model 8¶

Observations¶

Model 9¶

Observations¶

Model 10¶

Observations¶

Model 11 - reduce learning rate and early stopping¶

Observations¶

Model 12 - reduce learning rate¶

Observation¶

Model 13¶

Observation¶

Model Performance Comparison and Final Model Selection¶

Observation¶

Choosing Best Model reasoning¶

Actionable Insights and Recommendations¶

Actionable Insights¶

Business Recommendation¶